GitLab

Bug: http://b/32834638
Test: make sdk.
Change-Id: Iffc2649dc131c73758db8727912c83c078d92071

Bug: http://b/32834638
Test: run python scripts like report_sample.py.
Change-Id: Ia37e7b34babe7ba3df88b38ec6e6bc15241f50d8

Taken from branch aosp-master.

Bug: http://b/32834638
Test: None.

Change-Id: I6d456a932f8ab679da4c34f4113dc7414fe29d62

This reverts commit 266cf44b.

And fix error in the original patch.

Bug: http://b/32834638
Test: make x86-sdk and mips-sdk.

Change-Id: Ie9bc21ca461252f54c8f443747c8d9d1b5d33c02

This reverts commit f6bc6f2e.

Change-Id: Ic8a2405372498ef8251bcf4b680408001c164550

Build 32/64 bit simpleperf on target, simpleperf on host,
and libsimpleperf_report.so on host in sdk. So they can
be used in prebuilts/simpleperf.

Bug: http://b/32834638
Test: make sdk and win_sdk manually.

Change-Id: I1e55c9b1596aca3b982dc30422d29d830077289e

Test: no more exception failures
Bug: 33846158
Change-Id: Iaaa43affae3a42cf5b94bed84ab94197d3d4d2d1

libselinux is being used as shared lib by many other modules.
So using its shared lib will reduce total image size.

Size diff on angler build image is as follows.

runconuid: 19912 -> 11256 (-8656)

Test: building succeeded, and the image was tested on angler.
Bug: 33056637
Change-Id: I6272486fdacef06c30d25f1718651757d873eb1b

Bug: 31348667

Test: compiles
Change-Id: I31eb72e4e18dca68255a36915fcc980c928936d7

Test: Soong's build.ninja identical before/after (except GlobRules)
Change-Id: I541c6bd662bdddbeac6ca6a3ec517e09444fb232

Bug: http://b/32340274
Test: run simpleperf_unit_test.
Change-Id: Icb96d4778cf527720b24aed58699da05b29c1bd4

Originally we use per thread maps, and copy the maps from parent thread
when processing ForkRecord. But there are situations that ForkRecords
are lost, then the new thread is left having no maps, which leads to unknown
dso/symbols.
To fix this, we can use per process maps. Even if ForkRecords are lost,
we know which process a SampleRecord belongs to, and can use the process maps.
This is reasonable because in most cases, new threads in the same thread group
are created with both CLONE_THREAD and CLONE_VM.

Bug: http://b/33503165
Test: run simpleperf_unit_test.
Change-Id: If84b6cdac39cd6ad4e441ed6d75ed879b86d1830

Test: builds
Bug: 27948817
Change-Id: I74fb403ff6173f475190894104fd8b1b82523fb5

Bug: http://b/32834638
Test: run python annotate.py on some apps.
Change-Id: Ib89e80af495ace6239351888db974e8fb5090403

Bug: http://b/33558210
Test: run simpleperf_unit_test.
Change-Id: If5a1f7fc9c436a39776f80f1e52479f1cd7ff407

Bug: None
Test: The warning is gone.
Change-Id: Ifbe4550b372333f81bc0ff88b25687dfd0b64be9

When monitoring some threads/processes not forked by simpleperf,
check if these threads/processes exist regularly. So we can
stop profiling once all threads/processes exit.

Also handle SIGHUP signal, so we can finish profiling properly
when `adb shell simpleperf record xxx` is killed by Ctrl-C.

Add corresponding tests.

Bug: http://b/33558210
Test: run simpleperf_unit_test.
Change-Id: Ieae4d00d099dc1c7a0c51b0610dff43981bb642e

app_profiler.py follows the instruction of profiling an
app in README.md. Users can just set the configuration,
and let app_profiler.py do environment setup, profiling
and data collection work.

Bug: http://b/32834638
Test: run app_profiler.py on android M, N, O.
Change-Id: I2e08790396b02df32320eaa27b75d49108f4856f

1. Reduce columns used by callgraph by reducing spaces between
entries from 11 to 4. This helps to show more entries in the limited screen width.
2. Add [hit-in-function] line for each entry, so we know the percentage
of samples hit in the function and samples hit in the children functions.
3. Add --kallsyms option for report command.

After all, The callgraph is changed like below.
Before change:
 |--99.08%-- java.lang.Thread.run
 |           void com.example.cyb.tappydefender.TDView.run()
 |           |
 |           |           |--95.63%-- void com.example.cyb.tappydefender.TDView.draw()

After change:
 |--98.85%-- java.lang.Thread.run
 |           void com.example.cyb.tappydefender.TDView.run()
 |    |--0.09%-- [hit in function]
 |    |
 |    |--95.54%-- void com.example.cyb.tappydefender.TDView.draw()

Bug: http://b/33356538
Test: run simpleperf_unit_test.
Test: run simpleperf report -g manually.
Change-Id: Ifc1846d4ee7469cadbcc727cae6bf59552d919d5

Change-Id: I9e02cda707b551980683895ef7f22522157628d8

Add option to specify the number of inodes to create when
making the filesystem.  Read-only partitions don't have
a need for extra inodes, a lot of space is wasted by these
inodes.

Bug: 32246383
Test: make vendor partition with new options
Change-Id: I7063972ec4f4e617d28d2e029ad6a9f8e94add73
(cherry picked from commit 15c434682ac78262d2dad9730805eab52b748cdf)

am: a5a10129

Change-Id: I0c2a0c8037cd95e25dab555e9c348ab7270cfaf7