OSDN Git Service
Mathieu Chartier [Sat, 6 Aug 2016 01:56:54 +0000 (01:56 +0000)]
Merge "Don't use profiles with vm safe mode" into nyc-mr1-dev
Mathieu Chartier [Sat, 6 Aug 2016 00:48:12 +0000 (17:48 -0700)]
Don't use profiles with vm safe mode
vm safe mode uses interpret only compiler filter, this filter will
generate a full app image if a profile is also passed down. App
image generation without a profile currently has a bug where uses
library is not pruned correctly.
Bug:
30688277
Change-Id: I60d87d698789583b8ab797eb47d031c96caf6d9b
Test: N5X boot, package compile speed-profile comb.BBClient
Peng Xu [Fri, 5 Aug 2016 18:03:44 +0000 (18:03 +0000)]
Merge "Tune SensorService thread priority, lower sensor event latency" into nyc-mr1-dev
Pablo Ceballos [Fri, 5 Aug 2016 17:14:45 +0000 (17:14 +0000)]
Merge "SF: Change rounding behavior in setGeometry" into nyc-mr1-dev
Peng Xu [Tue, 2 Aug 2016 01:12:11 +0000 (18:12 -0700)]
Tune SensorService thread priority, lower sensor event latency
Change SensorService priority to SCHED_FIFO 10 to ensure
sensor events to be delivered to apps in consistent short time even
if system is under load.
Bug:
30105910
Change-Id: I5ddf7061b75cf46b54e16476d20bbc888024aa43
Felipe Leme [Thu, 4 Aug 2016 19:48:50 +0000 (12:48 -0700)]
Fixed logic to print BEGIN: line for bugreportz.
BUG:
30451114
Change-Id: I0e65a38999c128a435b43ad0806e06e6e74580ff
TreeHugger Robot [Thu, 4 Aug 2016 14:44:34 +0000 (14:44 +0000)]
Merge "Added a BEGIN:path message on bugreportz protocol." into nyc-mr1-dev
Pablo Ceballos [Wed, 3 Aug 2016 17:20:45 +0000 (10:20 -0700)]
SF: Change rounding behavior in setGeometry
- Add an option for a transform to round outwards instead of to the
nearest pixel.
- Use this new rounding behavior when converting the crop rectangle
from display-space to layer-space. Otherwise a pixel that will
partially be on the screen might be marked as transparent.
Bug
30510632
Change-Id: I1aece177cc29c54ee0d4a2c919e47442eb455f30
Tim Murray [Tue, 2 Aug 2016 22:43:59 +0000 (15:43 -0700)]
Disable SCHED_FIFO when turning off the display.
There can be long SF runtimes when turning the display on or off. Only
use SCHED_FIFO when we are actively going to be rendering frames, not
when turning the display on or off.
bug
30375418
Change-Id: Ibef8e5f3bba0bafc36f92f9f1b299f015fc335f9
Felipe Leme [Fri, 29 Jul 2016 16:49:04 +0000 (09:49 -0700)]
Added a BEGIN:path message on bugreportz protocol.
BUG:
30451114
Change-Id: I3607c75b184e71a9a5a6393bdbf68200abe0fc16
Tim Murray [Fri, 29 Jul 2016 20:57:24 +0000 (13:57 -0700)]
Update SurfaceFlinger FIFO configuration.
- move SurfaceFlinger main thread to prio 2
- drop EventThread from FIFO but not SFEventThread, which is still FIFO
prio 1
- make DispSyncThread FIFO, prio 1
bug
30481949
Change-Id: Ifedccfbf3cf144ffad030109849a924dddae8d09
Dan Stoza [Fri, 29 Jul 2016 18:34:39 +0000 (18:34 +0000)]
Merge "SF: Prime shader cache before boot animation" into nyc-mr1-dev
Dan Stoza [Thu, 28 Jul 2016 20:31:51 +0000 (13:31 -0700)]
SF: Prime shader cache before boot animation
Primes the shader cache before launching boot animation to avoid
janking when we eventually fall into GLES composition.
Bug:
30040263
Change-Id: I76ba66e92c5e6bb05930c325f384f3b774e91101
mukesh agrawal [Wed, 27 Jul 2016 19:01:49 +0000 (12:01 -0700)]
dumpsys: fix formatting of service dump times
The current formatting of service dump times is
confusing some bugreport parsing tools. This is
because the current service dump time message
is hard to distinguish from the bugreport section
marker.
Adjust the formatting, to make the service
dump time message distinct from the bugreport
section marker.
BUG=
30401031
TEST=manual
Manual test
$ adb bugreport /tmp/bugreport.zip
$ cd /tmp
$ mkdir br
$ mv bugreport.zip br
$ cd br
$ grep 'was the duration' bugreport*.txt
------ 4.283s was the duration of 'DUMPSYS MEMINFO' ------
------ 0.053s was the duration of 'DUMPSYS CPUINFO' ------
[...many lines omitted...]
--------- 0.003s was the duration of dumpsys AtCmdFwd
--------- 0.002s was the duration of dumpsys DockObserver
[...many more lines omitted...]
Change-Id: I9d3dd75210b3984240a4b7ab49a1cc6124a269d8
(cherry picked from commit
50b7e039500b3decc78d5ffff9919d1481df4964)
TreeHugger Robot [Wed, 27 Jul 2016 23:44:27 +0000 (23:44 +0000)]
Merge "Build SF executable with USE_HWC2 when needed" into nyc-mr1-dev
Dan Stoza [Wed, 27 Jul 2016 22:05:38 +0000 (15:05 -0700)]
Build SF executable with USE_HWC2 when needed
The SurfaceFlinger executable wasn't picking up the USE_HWC2 flag when
enabled by the TARGET_USES_HWC2 BoardConfig, causing a mismatch in the
size of the SurfaceFlinger object. This fixes that by also passing
USE_HWC2 when building the executable.
Bug:
30302693
Change-Id: I1e41e5f548d4b8329590a082d9b69155ba365be9
TreeHugger Robot [Wed, 27 Jul 2016 18:45:54 +0000 (18:45 +0000)]
Merge "HWC2: Clear transform for SolidColor layers" into nyc-mr1-dev
Dan Stoza [Wed, 27 Jul 2016 17:16:52 +0000 (10:16 -0700)]
HWC2: Clear transform for SolidColor layers
Clears the transform for SolidColor layers, since it doesn't make sense
to transform layers without a buffer.
Bug:
30429522
Change-Id: Ib2f4889531556c7ab00e3b270319c321e9f32c84
Felipe Leme [Fri, 22 Jul 2016 19:03:20 +0000 (12:03 -0700)]
Added a -p option to bugreportz to show progress.
BUG:
28609499
Change-Id: I1e60078dfda7e5679fbd19f4981a5dc7a69c4fc7
Felipe Leme [Fri, 22 Jul 2016 03:10:57 +0000 (20:10 -0700)]
Added unit tests to bugreportz.
BUG:
28609499
Change-Id: I5b846eeeaa7c05c3e3f66f36d31ef42c472a3099
Dan Stoza [Wed, 20 Jul 2016 19:52:38 +0000 (12:52 -0700)]
HWC2: Use SolidColor composition for DimLayer
Switches SurfaceFlinger to try to use SolidColor HWC device
composition for DimLayers instead of always falling back to client
composition.
Bug:
30216498
Change-Id: Ia8af95d214ba72b5657f32f7827a1e6278629f17
Eino-Ville Talvala [Fri, 22 Jul 2016 21:42:36 +0000 (21:42 +0000)]
Merge "BufferQueue consumers: Add discardFreeBuffer method" into nyc-mr1-dev
Pablo Ceballos [Fri, 22 Jul 2016 21:15:43 +0000 (21:15 +0000)]
Merge "SF: Make screenshot surface async" into nyc-mr1-dev
Dan Stoza [Fri, 22 Jul 2016 21:11:15 +0000 (21:11 +0000)]
Merge changes I7f7d4e0a,Ie966e37d into nyc-mr1-dev
* changes:
HWC2: Check all displays for client composition
HWC2: Kill logspam for non-HWC virtual displays
Eino-Ville Talvala [Fri, 22 Jul 2016 00:06:58 +0000 (17:06 -0700)]
BufferQueue consumers: Add discardFreeBuffer method
This method releases all free buffers owned by the buffer queue,
in order to save memory (at the cost of potential future
reallocation of buffers).
Bug:
28695173
Change-Id: I458d10373e639e3144faf673af2ba01aca36e65a
Jeff Sharkey [Fri, 22 Jul 2016 19:09:51 +0000 (19:09 +0000)]
Merge "Ignore setxattr() failures on tmpfs." into nyc-mr1-dev
Jeff Sharkey [Fri, 22 Jul 2016 17:38:54 +0000 (11:38 -0600)]
Ignore setxattr() failures on tmpfs.
Bug:
30210614
Change-Id: Ic9dfb2439b4b1056e1a58552be2e84f4e524321e
Dan Stoza [Thu, 21 Jul 2016 21:49:33 +0000 (14:49 -0700)]
HWC2: Check all displays for client composition
SurfaceFlinger currently only checks whether the primary display
performed client composition when considering whether to propagate
backpressure back to apps. This change expands that check to all
displays so that if an external or virtual display is using client
composition it disables backpressure propagation as well.
Bug:
30022738
Change-Id: I7f7d4e0a1ea7d27ef1f280e2b4ebd5e5bd3bd911
Pablo Ceballos [Thu, 21 Jul 2016 20:44:51 +0000 (13:44 -0700)]
SF: Make screenshot surface async
It will only ever have one frame pushed to it and this guarantees that
the late-latching logic will always latch it immediately.
Bug
30209608
Change-Id: I875cb5ea258ebb0afae5d1af4ad946475c2240bc
Dan Stoza [Thu, 21 Jul 2016 18:09:40 +0000 (11:09 -0700)]
HWC2: Kill logspam for non-HWC virtual displays
Removes some logspam and fixes some non-errors which were reported as
errors when running a virtual display that is not backed by hardware
composer.
Bug:
30022738
Change-Id: Ie966e37d95f53bcc050eef064d24bbe26748b7fd
TreeHugger Robot [Thu, 21 Jul 2016 05:27:24 +0000 (05:27 +0000)]
Merge "HWC2: Add properties to revert latching changes" into nyc-mr1-dev
Dan Stoza [Wed, 20 Jul 2016 22:38:12 +0000 (15:38 -0700)]
HWC2: Add properties to revert latching changes
Adds two properties:
debug.sf.latch_unsignaled - This causes SurfaceFlinger to latch
buffers even if their fences haven't signaled
debug.sf.disable_backpressure - This causes SurfaceFlinger to skip
propagating backpressure back to apps by not running when it
detects that a prior composition missed hardware vsync
Bug:
30216498
Change-Id: Idd70f472ec567a4bc03352f30f201f434cf5d660
TreeHugger Robot [Wed, 20 Jul 2016 17:21:33 +0000 (17:21 +0000)]
Merge "Color transforms are now color modes." into nyc-mr1-dev
Lajos Molnar [Tue, 19 Jul 2016 22:31:22 +0000 (15:31 -0700)]
OMX: document time limits for some operations
Bug:
28408062
Bug:
28407319
Change-Id: I7e26eac1d653c871f129b0610e65221b081ba040
Pablo Ceballos [Mon, 18 Jul 2016 22:35:16 +0000 (22:35 +0000)]
Region: Detect malicious overflow in unflatten am:
1a65fccc50 am:
fd579a51b5 am:
0e85f4d48d am:
0ec362e024
am:
b82c6f52c0
Change-Id: I2aed411154ae53dd27b5f8c76f477afe7136eca3
Pablo Ceballos [Mon, 18 Jul 2016 22:32:55 +0000 (22:32 +0000)]
Region: Detect malicious overflow in unflatten am:
1a65fccc50 am:
fd579a51b5 am:
0e85f4d48d
am:
0ec362e024
Change-Id: Ibf1f0aedeab8552fe2a211d01d8b5c2c1ddafcdd
Pablo Ceballos [Mon, 18 Jul 2016 22:30:46 +0000 (22:30 +0000)]
Region: Detect malicious overflow in unflatten am:
1a65fccc50 am:
fd579a51b5
am:
0e85f4d48d
Change-Id: I6d8ed780a6ec9d8e5b3ec17b20bce528835b4bf0
Pablo Ceballos [Mon, 18 Jul 2016 22:28:15 +0000 (22:28 +0000)]
Region: Detect malicious overflow in unflatten am:
1a65fccc50
am:
fd579a51b5
Change-Id: I2d584199f001f7da8f4331729d796cf527ba79fd
Pablo Ceballos [Mon, 18 Jul 2016 22:25:45 +0000 (22:25 +0000)]
Region: Detect malicious overflow in unflatten
am:
1a65fccc50
Change-Id: I597a130dfe82c21c5611f45f6977b189c0a1a8ea
Michael Wright [Tue, 12 Jul 2016 20:30:53 +0000 (13:30 -0700)]
Color transforms are now color modes.
Rename color transforms to color modes for all interfaces exposed to
surfaceflinger clients. Also split it out to be a separate
configuration value from display modes.
Bug:
29044347
Change-Id: I87e937f7c954a50c946e8e2c606797caa416c5d8
Jeff Sharkey [Fri, 15 Jul 2016 20:53:24 +0000 (20:53 +0000)]
Merge "Record "cache" inodes to clear while CE is locked." into nyc-mr1-dev
Dan Stoza [Wed, 22 Jun 2016 21:51:09 +0000 (14:51 -0700)]
SF/HWC2: Add support for color transforms
Adds support for color transforms using the setColorTransform method
of HWC2.
This means that instead of always falling back to client composition
when applying a transform, SurfaceFlinger will allow the device to
make that decision. If all layers fall back to client composition, the
SKIP_CLIENT_COLOR_TRANSFORM capability allows the device greater
control over whether SF should apply the transform or whether it
should allow the device to apply it to the client target buffer.
Bug:
19539930
Change-Id: I47a3d5453a3c47a8dd105ab77cce7f9c9687e925
Jeff Sharkey [Fri, 15 Jul 2016 00:16:22 +0000 (18:16 -0600)]
Record "cache" inodes to clear while CE is locked.
There are two situations where we need to clear cached data from
all users, including those whose CE storage might still be locked:
1. When PackageManager is clearing caches to try making room for
package updates.
2. When the device fingerprint changes, we need to clear code caches
for all apps.
To enable this, we now record the inode number of the "cache" and
"code_cache" directories in xattrs of the parent directory. This is
just enough information to find the cache directories to enable
deleting files inside. When preparing CE storage for an app, we now
create these two directories and immediately write the inode numbers.
Bug:
26056125
Change-Id: I7e442b0676a695acf962593469793a93b03c8aee
TreeHugger Robot [Thu, 14 Jul 2016 19:54:21 +0000 (19:54 +0000)]
Merge "HWC2: Fix virtual displays" into nyc-mr1-dev
TreeHugger Robot [Thu, 14 Jul 2016 19:49:07 +0000 (19:49 +0000)]
Merge "Disable support for EGL_ANDROID_get_frame_timestamps" into nyc-mr1-dev
Sergio Giro [Thu, 14 Jul 2016 18:07:54 +0000 (18:07 +0000)]
Add bound checks to utf16_to_utf8
am:
9b39ebe132
Change-Id: Ia3ce9b64333983799c6fdb3c5f7383852abda32e
Pablo Ceballos [Thu, 14 Jul 2016 15:02:14 +0000 (08:02 -0700)]
Disable support for EGL_ANDROID_get_frame_timestamps
Bug
22764477
Change-Id: Ic7abe534b2f712dd1442364eb1fec3d1fbe0ee91
Dan Stoza [Wed, 13 Jul 2016 22:48:58 +0000 (15:48 -0700)]
HWC2: Fix virtual displays
Close the output buffer release fence per the contract in
hwcomposer2.h.
Also changes the HWC2 device to hold a weak pointer to the display,
which allows it to properly remove it when it gets torn down.
Bug:
29945684
Change-Id: Ib1b5d81e42a63d0e65b189f7c6c62e661e0ab6ae
Pablo Ceballos [Wed, 13 Jul 2016 21:11:57 +0000 (14:11 -0700)]
Region: Detect malicious overflow in unflatten
Bug
29983260
Change-Id: Ib6e1cb8ae279010c5e9960aaa03513f55b7d873b
Sergio Giro [Tue, 28 Jun 2016 17:19:33 +0000 (18:19 +0100)]
Add bound checks to utf16_to_utf8
Bug:
29250543
Change-Id: I518e7b2fe10aaa3f1c1987586a09b1110aff7e1a
(cherry picked from commit
7e93b2ddcb49b5365fbe1dab134ffb38e6f1c719)
Lajos Molnar [Thu, 14 Jul 2016 01:54:14 +0000 (01:54 +0000)]
Merge "media: add asString for new OMX layering params" into nyc-mr1-dev
Wei Liu [Wed, 13 Jul 2016 23:59:27 +0000 (23:59 +0000)]
Merge "Add bugreportwear service for Wear." into nyc-mr1-dev
Lajos Molnar [Wed, 13 Jul 2016 23:31:30 +0000 (16:31 -0700)]
media: add asString for new OMX layering params
Bug:
27596987
Change-Id: I75af5f8a92cf00a43bfa209793db3e79f8621374
Lajos Molnar [Wed, 13 Jul 2016 23:28:06 +0000 (23:28 +0000)]
Merge "media: add/fix asString for temporal layering types" into nyc-mr1-dev
Andreas Gampe [Tue, 12 Jul 2016 16:47:17 +0000 (09:47 -0700)]
Otapreopt: Try to mount vendor partition for A/B OTA
Vendor apps are usually not preopted, so A/B dexopt should pick
them up. update_engine is not mounting the vendor partition, so
let otapreopt_chroot do the work.
Bug:
25612095
Bug:
29498238
Change-Id: I5a77bdb78a8e478ce10f6c1d0f911a8d6686becb
TreeHugger Robot [Wed, 13 Jul 2016 18:57:20 +0000 (18:57 +0000)]
Merge "media: clarify i-frame interval for VPx and HEVC codecs" into nyc-mr1-dev
Andreas Gampe [Wed, 13 Jul 2016 18:29:47 +0000 (18:29 +0000)]
Merge changes from topic 'ab_ota_dexopt_target_slot' into nyc-mr1-dev
* changes:
Otapreopt: Add init script for A/B artifact move
Otapreopt: Implement new A/B OTA path
Andreas Gampe [Sat, 9 Jul 2016 18:44:12 +0000 (11:44 -0700)]
Otapreopt: Add init script for A/B artifact move
Move the /data/ota -> /data/dalvik-cache move from the zygote into
a minimal shell script. Add an init rc file to execute the script
once /data is mounted. Add Makefile rules to pull these files in
with the rest of the A/B OTA dexopt components.
The move is motivated by the new naming scheme that includes the
slot name. Passing the slot name to the zygote would complicate
the process, while the value is immediately available in the script
through a simple getprop call.
Bug:
25612095
Bug:
28069686
Change-Id: I63e718d488662e1b1de2ce857629cb90aa4f611d
Andreas Gampe [Mon, 27 Jun 2016 21:25:30 +0000 (14:25 -0700)]
Otapreopt: Implement new A/B OTA path
Refactor otapreopt_script, otapreopt_chroot and otapreopt such
that the actual work will be done as a child of the script driven
by update_engine.
The script now uses the new OtaDexoptService command to get the
right parameters for a dexopt call in otapreopt. As we reach
otapreopt_chroot and otapreopt directly, we can add parameters
without tainting the regular installd path. Use this to add the
target slot suffix, and make naming of the A/B artifacts include
the suffix instead of the generic "b."
Otapreopt_chroot startup is slightly complicated because all file
descriptors must be closed. Otherwise, dex2oat needs selinux
permissions to inherit the descriptors from update_engine.
Bug:
25612095
Bug:
28069686
Change-Id: I9d7b86ac8ecfdf91af3800f7e96f41534c0afc0f
Lajos Molnar [Wed, 13 Jul 2016 01:39:09 +0000 (18:39 -0700)]
media: add/fix asString for temporal layering types
Bug:
27596987
Change-Id: I20c628a2f0ca9770351f7096ce05a2c383e53d41
Courtney Goeltzenleuchter [Tue, 12 Jul 2016 23:12:00 +0000 (23:12 +0000)]
Merge changes Iea048eaa,I5c246571 into nyc-mr1-dev
* changes:
Add support for multiple color modes
Remove unused DisplayInfo structure
Praveen Chavan [Fri, 25 Mar 2016 08:17:32 +0000 (01:17 -0700)]
media: add OMX extensions for video temporal layering
Add static/dynamic configuration structures to setup/tune
video encoder for temporal layering
Bug:
27596987
Change-Id: Ic260872aa49faee6d17a84e6177ef94c35ef0d68
Lajos Molnar [Tue, 12 Jul 2016 20:59:47 +0000 (13:59 -0700)]
media: clarify i-frame interval for VPx and HEVC codecs
Bug:
30096321
Change-Id: I421487e6ba3d4054f9a179dc5109a01564b945f4
TreeHugger Robot [Tue, 12 Jul 2016 17:51:11 +0000 (17:51 +0000)]
Merge "Installd: Add support for dex2oat very-large-app property" into nyc-mr1-dev
Andreas Gampe [Mon, 11 Jul 2016 22:19:31 +0000 (15:19 -0700)]
Installd: Add support for dex2oat very-large-app property
To support constrained devices, dex2oat supports a parameter that
can downgrade a compilation to verify-at-runtime. Expose this
through the dalvik.vm.dex2oat-very-large system property.
Bug:
29557002
Change-Id: Iace626a8ec18533d37f84a41c20729fe7e400386
Pablo Ceballos [Tue, 12 Jul 2016 00:24:56 +0000 (00:24 +0000)]
Merge "BQ: Remove getNextFrameNumber Binder call" into nyc-mr1-dev
Pablo Ceballos [Tue, 12 Jul 2016 00:17:48 +0000 (00:17 +0000)]
Merge "BQ: Handle errors in GraphicBuffer/Fence unflatten" into nyc-mr1-dev
Courtney Goeltzenleuchter [Thu, 23 Jun 2016 17:49:50 +0000 (11:49 -0600)]
Add support for multiple color modes
Bug:
29044347
Change-Id: Iea048eaa62f072a9bbefc4f3a6c29a9e593eab69
Courtney Goeltzenleuchter [Tue, 21 Jun 2016 23:22:18 +0000 (17:22 -0600)]
Remove unused DisplayInfo structure
Bug:
29044347
Change-Id: I5c2465719050c8727b0b48ef515ee79a24e3d65d
Pablo Ceballos [Mon, 11 Jul 2016 19:16:48 +0000 (19:16 +0000)]
Merge "Add the EGL_ANDROID_get_frame_timestamps extension" into nyc-mr1-dev
smain@google.com [Mon, 11 Jul 2016 17:05:06 +0000 (17:05 +0000)]
update NDK doxygen templates to create DevSite-style HTML.
am:
20e2095c7b
Change-Id: I0ae956648ff553bbf15c083175451621555fdf38
Sergio Giro [Mon, 11 Jul 2016 12:11:30 +0000 (12:11 +0000)]
Merge "Add bound checks to utf16_to_utf8" into nyc-mr1-dev
Wei Liu [Fri, 8 Jul 2016 23:42:01 +0000 (16:42 -0700)]
Add bugreportwear service for Wear.
b/
30041885
Change-Id: Ic054e4406cbf067b86ea79aee59a105b82344fec
TreeHugger Robot [Fri, 8 Jul 2016 21:22:43 +0000 (21:22 +0000)]
Merge "HWC2: Backpressure on transactions as well" into nyc-mr1-dev
Dan Stoza [Fri, 8 Jul 2016 19:02:20 +0000 (12:02 -0700)]
HWC2: Backpressure on transactions as well
Modifies the backpressure detection to skip transactions as well as
buffer updates.
Bug:
29413700
Change-Id: I97cda920e7fc1cd7151f0a4df1b00f96b84f4145
Pablo Ceballos [Tue, 31 May 2016 21:55:42 +0000 (14:55 -0700)]
Add the EGL_ANDROID_get_frame_timestamps extension
Change-Id: Ia7d1c10f0b8bd1f2f6dc7dc180764cb1b4fdbf6f
Peng Xu [Fri, 8 Jul 2016 17:40:33 +0000 (17:40 +0000)]
Merge "Increase sensor registration log size" into nyc-mr1-dev
smain@google.com [Fri, 8 Jul 2016 00:11:35 +0000 (17:11 -0700)]
update NDK doxygen templates to create DevSite-style HTML.
Previously, we had doxygen create .jd files with some syntax that
was parsed by the DocLava build tools when creating the final HTML
files for the website. But now that we're publishing these docs through
Piper, we don't need to run them through DocLava anymore. So with this
change, we now create .html files that then need to be copied to
google3/googledata/devsite/site-android/en/ndk/
where they can be immediately published via devsite once merged.
bug:
29943712
Change-Id: I8af5555f802e8c0122378bd504ad4308ff9aad56
Peng Xu [Thu, 7 Jul 2016 22:40:08 +0000 (15:40 -0700)]
Increase sensor registration log size
This is for aiding debugging of b/
29625330. The size is increased
so that we have a better chance to know what happened before the
failure happens.
Bug:
29625330
Change-Id: I8b741cb5981d8e91b9ef53a00e65c0bbe34ed55e
Tim Murray [Wed, 22 Jun 2016 19:42:10 +0000 (12:42 -0700)]
surfaceflinger: make critical threads SCHED_FIFO
Sets the main thread, EventThread, and SFEventThread to SCHED_FIFO to
minimize jitter.
bug
24503801
Change-Id: I3751c00c79482842f5c0bdaec92b6cbfe62328a0
Pablo Ceballos [Wed, 6 Jul 2016 22:24:54 +0000 (15:24 -0700)]
BQ: Handle errors in GraphicBuffer/Fence unflatten
Bug
29873396
Change-Id: Ieab38091f60722a11e227aecac7508fc11d9f58b
Dan Stoza [Fri, 1 Jul 2016 21:00:40 +0000 (14:00 -0700)]
Surface: Add LAST_[DE]QUEUE_DURATION queries
Adds support for the NATIVE_WINDOW_LAST_DEQUEUE_DURATION and
NATIVE_WINDOW_LAST_QUEUE_DURATION queries.
Bug:
29413700
Change-Id: Iea4efa208e8390fb2de52d5ad9f083ae91bf50b0
Dan Stoza [Fri, 1 Jul 2016 20:33:38 +0000 (13:33 -0700)]
HWC2: Backpressure missed vsyncs into apps
Adds a mechanism to detect whether the prior frame SurfaceFlinger
presented to hardware composer was actually picked up or not. We then
use this mechanism to avoid pushing two frames in the same vsync.
This backpressure is passed back to applications by not latching any
buffers, which will manifest as dequeueBuffer stalling until
SurfaceFlinger releases a buffer on the following vsync.
Also makes the former INVALIDATE_ON_VSYNC behavior the only behavior
so that this functionality works correctly.
Bug:
29413700
Change-Id: Ibde358e45423ee6fea7b5e09ff65e49c4ad67baa
Dan Stoza [Wed, 15 Jun 2016 23:26:27 +0000 (16:26 -0700)]
SF: Only latch buffers after fence signals
Changes SurfaceFlinger to only latch a buffer after its corresponding
acquire fence has signaled. This will enable us to move
SurfaceFlinger closer to vsync since there is no risk of fence waits
blocking composition.
Bug:
29413700
Change-Id: I26f4fd600c1611b8d736ec654d1f0f02cf69ae5f
Pablo Ceballos [Fri, 1 Jul 2016 21:15:41 +0000 (14:15 -0700)]
BQ: Remove getNextFrameNumber Binder call
- Return the value in queueBuffer instead and cache it in Surface
Change-Id: I10ab112afb03cf0231b047d4a4569cd641827043
Pablo Ceballos [Thu, 30 Jun 2016 23:48:02 +0000 (16:48 -0700)]
resolve merge conflicts of
8e3e92b to nyc-mr1-dev
Change-Id: Ic67f0894e68ea5276eb9187a13cd24f105d55d68
Sergio Giro [Tue, 28 Jun 2016 17:19:33 +0000 (18:19 +0100)]
Add bound checks to utf16_to_utf8
Bug:
29250543
Change-Id: I518e7b2fe10aaa3f1c1987586a09b1110aff7e1a
Pablo Ceballos [Wed, 29 Jun 2016 22:08:33 +0000 (15:08 -0700)]
BQ: Handle buffer allocation failure
Return the slot to the mFreeSlots if we fail to allocate.
Bug
29873396
Change-Id: Icafc5c64c7233d47e160f7a56e5a314e4a01611d
mukesh agrawal [Thu, 9 Jun 2016 01:16:36 +0000 (18:16 -0700)]
dumpsys: report per-service dump times
In WiFi soak testing, we've seen several cases where
dumpsys times out. We've been unable to pin down the
cause of the timeouts, as there's no single service
that is obviously hung.
Add reporting of service dump times, to help figure
out why dumpsys is timing out.
BUG=
29090949
TEST=manual
Manual test
- load build on bullhead
$ adb shell dumpsys | grep 'was the duration'
- see non-zero values
Change-Id: Iaba60d840ac55cba171a3ccbc74c2aee8b23dce6
(cherry picked from commit
4e4c548437b7cc7748e97ecc74a525e26a441cdb)
Wei Wang [Tue, 28 Jun 2016 21:32:35 +0000 (14:32 -0700)]
dumpstate: change OOM adjustment proc interface for modern kernel
Bug:
29831067
Change-Id: I72e2862daad5da99e7d767d7e41bde96750840ce
Pablo Ceballos [Tue, 28 Jun 2016 00:56:53 +0000 (17:56 -0700)]
BQ: Add and expose a unique id
Bug
29422927
Change-Id: I80eab94f073ebc378302f00fa86a740c3643657e
Mark Salyzyn [Fri, 24 Jun 2016 23:31:24 +0000 (23:31 +0000)]
dumpstate: Also pick up /sys/fs/pstore/console-ramoops-0 for LAST KMSG
am:
ced6078b0e
Change-Id: Iacabdb52ae0e4ad3a6c27e952fa9ce572b282904
Mark Salyzyn [Fri, 24 Jun 2016 21:06:15 +0000 (14:06 -0700)]
dumpstate: Also pick up /sys/fs/pstore/console-ramoops-0 for LAST KMSG
(cherry pick from commit
7d0a762ecaf9f4d005f0f6db913034c2e084d362)
Bug:
29639467
Change-Id: Iba0b1754eaebfb2b0624dcbef522b7d8456eeb50
Mark Salyzyn [Mon, 16 May 2016 15:33:59 +0000 (08:33 -0700)]
dumpstate: sscanf requires an asciiz string
(cherry pick from commit
290f4b97bb736b0652eac9667965444393e683ec)
Bug:
28770668
Change-Id: Ice56b129a243296b2691d55e8f8c35b478b1f481
Greg Kaiser [Thu, 23 Jun 2016 03:31:59 +0000 (03:31 +0000)]
Merge \"Sensor: Make getId() more varied\" into nyc-dev
am:
c0ca0ceaa5
Change-Id: I6b6ecc30646e6e9a912a1ad096bd93fea341968b
Greg Kaiser [Thu, 23 Jun 2016 03:25:44 +0000 (03:25 +0000)]
Merge "Sensor: Make getId() more varied" into nyc-dev
Greg Kaiser [Tue, 21 Jun 2016 23:11:14 +0000 (16:11 -0700)]
Sensor: Make getId() more varied
Apps from different developers will now receive a different
ID for the same dynamic sensor. Additionally, all apps
will now receive a different/new ID for the same dynamic
sensor after a factory reset.
Bug:
28775590,
29547335
Change-Id: I368655b0e4bdc16736a7fd163ea3f7dad2886b3b
Jim Miller [Wed, 22 Jun 2016 22:58:31 +0000 (15:58 -0700)]
Enable fingerprint sensor navigation to invoke QuickSettings
- rename "FP_NAV" to more generic "SYSTEM_NAVIGATION" based on other reviews.
Fixes bug
28869612
Change-Id: I384c0fa372a2ecda7250fccd0dbdfd6312423a85
Felipe Leme [Wed, 22 Jun 2016 20:29:24 +0000 (20:29 +0000)]
Merge \"Run iptables as root.\" into nyc-dev
am:
ab69cc1f11
Change-Id: I4ca2677e7b99cfaa1ccc60c3efb2df642f3a7408
TreeHugger Robot [Wed, 22 Jun 2016 20:24:53 +0000 (20:24 +0000)]
Merge "Run iptables as root." into nyc-dev