It is possible for a match to only contain a context and not a schema,
or vice versa and in this case, the previously matched values should
continue to be used. The serializer already handles this case by
setting the index of the match to ~0u if there is not one, so this
change simply has the parser skip these values.
Bug: 70858511
Test: unit tests, including new ones
Change-Id: Ibc65bd8d637e39f3b1ce7bcc2b88189b75173f88
Add a benchmark to measure how long we pause a process when dumping.
Bug: http://b/62112103
Test: manually ran it
Change-Id: Iceec2f722915b0ae26144c86dcbeb35793f963da
Partners (entities other than AOSP) can expose their own system
libraries which are in /system/lib[64] to Android apps. This can be done
by adding the name of the libs into the files
/system/etc/public.libraries-<companyname>.txt.
There can be multiple of the txt files on a device, which is for the
case that multiple partners contributing to the same system image
have their own set of public libraries.
The public libraries MUST be named as lib<name>.<companyname>.so. This
is to prevent accidental exposure of AOSP-defined system private libs.
Note 1:
<companyname> doesn't need to be the same as the value of the sysprop
ro.product.manufacturer or anything that can be part of a file path.
Note 2:
This feature is not for exposing SoC-specific libs to Android apps. That
is already done by /vendor/etc/public.libraries.txt and is only for
libs in /vendor/lib[64].
Bug: 68280171
Test: mm -j under /system/core/libnativeloader/test
cts-tradefed run commandAndExit cts-dev -m CtsJniTestCases -t
android.jni.cts.JniStaticTest#test_linker_namespaces
Change-Id: I7d32ec27c7858e35b96c7c98223cc110acb35b81
Shared libraries in the directory should be dlopened with full file
paths. That was a workaround for some legacy prebuilt binaries and they
are all now fixed. Thus removing.
Bug: 70551668
Test: walleye, aosp_walleye, taimen, aosp_taimen boot to the UI
Change-Id: Ifb123b09c13f873cbefa7784e76dfe5f35575a8d
1. Extend MAX_BACKTRACE_FRAMES to 256, as we can have callchains
with length near 256 when recording Android applications.
2. .eh_frame and .ARM.exidx doesn't know how to unwind from instructions
setting up or destroying stack frames. It can lead to wrong callchains,
which may contain pcs outside executable mapping areas. Stop unwinding
once this is detected.
3. Some CIE entries in .eh_frame/.debug_frame suggest unwinder reading
personality routine from memory. However, that is not available in
offline unwinding, and leads to ACCESS_MEM_FAILED error. Work around
this by returning all zero data when failed to read data for some
addresses.
4. Some libraries have empty .eh_frame_hdr but valid .eh_frame.
Work around this by building eh_frame_hdr manually.
5. Prefer to use .debug_frame than .eh_frame. Because the former
can unwind instructions setting up and destroying stack frames.
Bug: http://b/69383534
Test: run backtrace_test.
Test: run simpleperf on some apps.
Change-Id: I62dddda3865dd567c6be150ce5527dc2d3516256
Providing alternative suggestions for using C++ stdlib types
instead of libutils types:
- higher interoperability
- fewer "legacy" quirks
- ability to use stl algorithms
- high optimization levels
Test: none
Change-Id: If81aa9982ca0ad229fa13c8142387906981b054d
Reduce the amount of time that a process remains paused by pausing its
threads, fetching their registers, and then performing unwinding on a
copy of its address space. This also works around a kernel change
that's in 4.9 that prevents ptrace from reading memory of processes
that we don't have immediate permissions to ptrace (even if we
previously ptraced them).
Bug: http://b/62112103
Bug: http://b/63989615
Test: treehugger
Change-Id: I7b9cc5dd8f54a354bc61f1bda0d2b7a8a55733c4
This is to help debugging different offline unwiding failures.
Bug: http://b/69383534
Test: run backtrace_test.
Change-Id: I5ed4837027a9f17d032925e97e9f5927161444b3
- Add a load_bias field in MapInfo so that it can be loaded offline,
and also so it can be cached.
- Add an Add function to the Maps class so that it's possible to manually
create a map.
- Remove the OfflineMaps class since I haven't found a reason for this to
exist.
- Add a pointer to the gnu debugdata compressed section in the interface
itself and modify the step path to try eh_frame, then debug_frame, then
gnu_debugdata. This way arm can add exidx as the last step behind
gnu_debugdata. Add an offline test to verify the order of unwind.
- Fix x86_64_ucontext_t since it was a different size on 32 bit and 64 bit
systems.
Test: Pass new unit tests.
Change-Id: I978b70d6c244bd307c62a29886d24c1a8cb2af23
The checksum is unnecessary. Improves adb performance by 40% on USB2.
Test: new adb works with new + old adbd, old adb works with new adbd
bug 67327728
Change-Id: I761d8a5a62deaea9bbb092ea9926b2d6d312f00d
Paths for extended VNDK libs don't need version suffix, because there
always is the single version that the vendor(odm) modules are built
against.
Bug: 70601582
Test: walleye boots to the UI.
Test: No further test is possible since we don't have extended VNDK yet
in our source tree.
Change-Id: Idbf4bb820ddb136d00744f64ddf6ebe6442ad16e
Re-format the files by splitting lines using +=.
Also add /odm/${LIB} where needed.
A few directories that are no longer required are removed from the
search paths and permitted paths.
Test: walleye and sailfish boots to the UI, renderscript, camera,
camcodder, sound, etc. are working.
Change-Id: I3150f0c3d35130d6b1a665e3f0813d33b1b7f546
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
Currently, we only check if the name of a given node in the trie is a
prefix match and the prefix matches array of the final node. This is
incorrect however, as the prefix matches array of intermediate nodes
may contain the prefix for a given property. This change adds that
check and test cases for this case.
Bug: 36001741
Test: new unit tests
Change-Id: I9f58ebc559f2ac591aa44df9e71205704bf18f66
Tom Cherry