adbd's file sync service doesn't handle a full socket gracefully,
immediately terminating the service as soon as it fails to write a
response. This would generally be fine if the socket's buffer were as
large as it claims (212992 by default with a 64-bit kernel), but this
buffer size is a giant lie, as each write has 576 bytes of overhead
that's used up in the send buffer. When setting the send buffer size,
the kernel helpfully doubles the value to attempt to account for the
overhead, but when writing 8 byte responses, only 2% of the buffer
actually gets used for responses, so we run out of buffer after 364
files instead of the 26624 that would be expected.
Fix this by processing the responses as they become available, and
calculate a maximum limit to how many sends we dispatch before we stop
and wait for responses to come in.
Bug: http://b/150827486
Test: manually modified adbd to respond with giant error messages, and
modified adb to not read responses until we choose to block
Change-Id: Ieb8c935662864211e2fd16c337ffed0992990086
(cherry picked from commit 672cdfeeff)
Make it so that we can get the sizeof a member of syncmsg without having
an instance of syncmsg or doing something awful along the lines of
sizeof(reinterpret_cast<syncmsg*>(nullptr)->status).
Bug: http://b/150827486
Test: m adb adbd
Change-Id: I4830e7f90033c7706ff52cdd8d13e9cf40c73628
(cherry picked from commit eddae92928)
The DexFile handle is allocated from heap in OpenFromFd/OpenFromMemory.
After releasing the unique_ptr, the DexFile handle itself is no longer
managed by the smart pointer. However, the DexFile handle is not freed
in the constructor of DexFileFromFile/DexFileFromMemory.
This change uses get() method to get the DexFile pointer while allowing
it to be managed by smart pointer so that it can be freed after method
end.
Added new unit tests to detect leaks.
Bug: 151966190
Test: Unwinding can still retrieve dex frame information during crash.
Test: Ran new unit tests before change and verified they fail, ran them
Test: after the change and verified they don't fail.
Signed-off-by: Yong Li <yongl0722@gmail.com>
Change-Id: I0627e1e255eb6644aba51e940c1a79ff78d568d7
(cherry picked from commit 489c3a8b35)
Some services are lazy HALs on some platforms and not lazy HALs on
others; this is known at runtime by hwservicemanager, so this change
adds these properties to allow hwservicemanager to turn one oneshot
(for lazy HALs). It may also be required to make a lazy HAL not lazy
anymore, and oneshot_off is provided for this.
Bug: 147841742
Test: new unit test that turn on and off oneshot on a service (bootanim)
and observes that it follows the expected behavior
Merged-In: I79524e2c9a5008f90c8d3bc40920fde00602a439
Change-Id: I79524e2c9a5008f90c8d3bc40920fde00602a439
(cherry picked from commit bdbf5047c9)
The ABI dump for libstatspull can't be built due to the link error.
Test: development/vndk/tools/header-checker/utils/create_reference_dumps.py
Bug: 151102177
Bug: 151694054
Change-Id: Id324890235679b9593995b6e12a60ef055b4c825
Print not more often than once a 100ms - it is smooth enough
and speeds up transfer even more on Windows, where a single
line output may take up to 5ms.
An added benefit is getting rid of some extra heap allocation
and string formatting when in the end the identical message
filtering would've dropped the line anyway. This is also
significantly more expensive on Windows.
Bug: 151900478
Test: manual, push/pull a file and a directory
Change-Id: I9038729e8a01d5f93fd301beaeb8a086f5039b77
1. Rename registerPullAtomCallback to setPullAtomCallback
2. Rename unregisterPullAtomCallback to clearPullAtomCallback
3. Change Ns to Millis for consistency with java
4. Swap metadata and callback params in setPullAtomCallback to match
java
5. Added getters for PullAtomMetadata
6. Added libstatspull_test to test PullAtomMetadata
7. Changed the types on setAdditiveFields from int to int32_t
Test: make
Test: atest LibStatsPullTests
Test: bit libstatspull_test:*
Bug: 1507885621
Bug: 151875223
Change-Id: I5cb07bfe71b1002180403828d5e8e6a5b88ce6fe
Support v3 header format when changing the command line with
fastboot boot or using flash:raw.
Bug: 151750405
Test: fastboot boot and flash:raw with updated cmdline and v3 header
Merged-In: Ibf396e2d18d8b22cad50db290f3fd4e46ff85d9b
Change-Id: Ibf396e2d18d8b22cad50db290f3fd4e46ff85d9b
Also reordered the directories in alphabetical order
Test: m -j
Bug: 149838525
Merged-In: I1a918d189d8bcb394ec6f818d033dbf7e4518713
Change-Id: I4b10a01ed6b3285aec6d87765f225c41ec55be96
Windows console IO is terribly slow. Reducing the number of
printed progress messages speeds up the transfer rate
from 80 to 130 MB/s on Windows laptop
Bug: 151900478
Test: adb push/pull
Change-Id: I223284c8a662bd8f2b8ba280cdcc8c930d3e5205
- Use one fewer heap allocation per fdevent object
- Lazy-init the fdevent context
Bug: 151239696
Test: various adb commands on Win/Linux
Change-Id: Ic7de207b30495e618f187e097c0276ad42c34005
Use only the syscalls that work with the wrapped ADB fds, or
extract the native handles for the case when need to call one
not wrapped.
Bug: 151239696
Test: adb install --incremental <apk> on Windows
Change-Id: Ia6de620171ab696b8136dcb60a2b63af6f86419f
It is possible for storaged to terminate due to SIGPIPE if its clients
have terminated. This is not a fatal condition from storaged's
perspective, so we ignore SIGPIPE instead.
Bug: 151581751
Test: build
Merged-In: I633780ed20908be02e5e1dea9504e140932afc8c
Change-Id: I633780ed20908be02e5e1dea9504e140932afc8c
(cherry picked from commit 3f3b1708d9)
To allow the ResourcesLoader API to load part of a file as an APK
that contains resources, an additional override of OpenArchiveFd
that contains read offset and length as parameters must be created.
This functionality allows for an APK stored in a zip file to be read
without having to write the APK to disk.
Bug: 142716192
Test: atest FrameworksResourceLoaderTests
Change-Id: I772fc8b462d71de0529717c420ced552103a6e3f
Merged-In: I772fc8b462d71de0529717c420ced552103a6e3f
We want to ignore SIGPIPE within init, but if we use SIG_IGN, that
would be inherited by child processes through exec(), which we do not
want to have happen. We instead set up a real signal handler with a
no-op handler function, that will ignore SIGPIPE within init, but will
not be inherited across exec().
This fixes c29c2baa69 ("init: Add support for native service
registration with lmkd"), when SIG_IGN was introduced.
Note that we caught this issue before shipping a release with that
change, so the major motivation here is to not cause a behavior change
in init.
Bug: 151581751
Test: children of init that don't explicitly block SIGPIPE exit when
sent SIGPIPE
Test: children of init that do explicitly block SIGPIPE do not exit
when sent SIGPIPE
Test: init does not exit when sent SIGPIPE
Test: init exits when sent SIGABRT
Merged-In: Ieda8555fd03836bcd672a422fe673a8369ad9beb
Change-Id: Ieda8555fd03836bcd672a422fe673a8369ad9beb
(cherry picked from commit fd470e87cc)
servingComplete_ was left uninitialized and only set to 'true'
in the code. We initialize it to the 'false' state to avoid
uninitialized references in SkipToRequest().
Bug: 150865433
Test: TreeHugger
Change-Id: Ia8a4d7135c432eb657543c5498fc9dbe8f4718b6
Before this change, "Success" is returned after all data is streamed,
around 7.5 seconds for Megacity.
After this change, "Success" is returned in about 1.5 seconds, before
streaming finishes.
BUG: 151676293
Test: manual
Change-Id: Ifda7de48da8e82623c99ae0194f70cb162fd72fa
Currently the server often quits before installation finishes. As a
result, there is no difference in the commandline output between a
successful installation and a failed one.
Let adb client wait till installation fails or succeeds by parsing the
output from the inc-server process.
Test: $ adb install --incremental ~/Downloads/base.apk
Test: Performing Incremental Install
Test: Serving...
Test: All files should be loaded. Notifying the device.
Test: Failure [INSTALL_PARSE_FAILED_NOT_APK: Failed to parse /data/app/vmdl749343150.tmp/base.apk: Failed to load asset path /data/app/vmdl749343150.tmp/base.apk]
Test: Install command complete (ms: 91 total, 0 apk prep, 91 install)
BUG: b/150865433
Change-Id: Ie33505f9cc08fc6d60ad4a5d709526e7aa9a0ad1
libbase doesn't have to rely on dlopen/dlsym to use liblog's new symbols
when it is built for __ANDROID_SDK_VERSION__ > 29.
Bug: 150860940
Test: TARGET_BUILD_APPS="com.android.adbd com.android.resolv" m
objdump -T ...shared_com.android.resolv/libbase.so | grep LIBLOG_R
=> should be none because resolv apex is targeting 29
objdump -T ...shared_com.android.adbd/libbase.so | grep LIBLOG_R
=> should list all new symbols because adbd apex is targeting R
objdump -T ...shared/libbase.so | grep LIBLOG_R
=> should list all new symbols
Change-Id: I7f7f16510d7637cd380fe35ea9ff3e804f38851d
1) Rename __android_logger_data to __android_log_message and rename
__android_log_write_logger_data to
__android_log_write_log_message. Move the const char* message
argument into __android_log_message.
2) Add @param, @return, and "Available since API level 30." to the
documentation of new functions.
3) Document that the user defined aborter should but may not abort.
4) Document the line separation is the responsibility of the log
function provided to __android_log_set_logger().
Bug: 150898477
Test: build, liblog and libbase unit tests
Change-Id: I07c41011ef25b3e7cc4943f3f1e240a2f6aa2802
Since new files can be introduced to /metadata after the initial
restorecon, we need to skip hashing.
Bug: 150907679
Test: rollback, ls -lZ /metadata/ota
Change-Id: I1f928f519ae2c75aeedb74b70d513743e86bcf2b
Merged-In: I1f928f519ae2c75aeedb74b70d513743e86bcf2b
A previous change moved property_service into its own thread, since
there was otherwise a deadlock whenever a process called by init would
try to set a property. This new thread, however, would send a message
via a blocking socket to init for each property that it received,
since init may need to take action depending on which property it is.
Unfortunately, this means that the deadlock is still possible, the
only difference is the socket's buffer must be filled before init deadlocks.
This change, therefore, adds the following:
1) A lock for instructing init to reboot
2) A lock for waiting on properties
3) A lock for queueing new properties
A previous version of this change was reverted and added locks around
all service operations and allowed the property thread to spawn
services directly. This was complex due to the fact that this code
was not designed to be multi-threaded. It was reverted due to
apparent issues during reboot. This change keeps a queue of processes
pending control messages, which it will then handle in the future. It
is less flexible but safer.
Bug: 146877356
Bug: 148236233
Bug: 150863651
Bug: 151251827
Test: multiple reboot tests, safely restarting hwservicemanager
Merged-In: Ice773436e85d3bf636bb0a892f3f6002bdf996b6
Change-Id: Ice773436e85d3bf636bb0a892f3f6002bdf996b6
(cherry picked from commit 802864c782)
This is apparently causing problems with reboot.
This reverts commit d2dab830d3.
Bug: 150863651
Test: build
Merged-In: Ib8a4835cdc8358a54c7acdebc5c95038963a0419
Change-Id: Ib8a4835cdc8358a54c7acdebc5c95038963a0419
create new profiles to allowing thawing and freezing back the freezer
cgroup
Bug: 151225245
Test: Manually verified that using the SetTaskProfiles method on the
profiles thaws and freezes back the freezer cgroup.
Change-Id: I7f3e193ebe79b49c1f6ac52b6138ff4ec26fc570
