This is a follow up CL for the previous commit [1], which may have
triggered an unknown bug in either Android Framework or LatinIME.
[1]: Id4d332e3909590c68345e10e1f2e18650efb2eb7
7b739a802cb7d97460a0bf8a1e9774efe96fa552
InputMethodService#mSettingsObserver is initialized in #onCreate() and
cleard with null in #onDestroy(). Hence hitting NPE against it implies
that InputMethodService#onEvaluateInputViewShown() can be called before
InputMethodService#onCreate() or after InputMethodService#onDestroy().
Both possibilities are equaly problematic. Note that this might be a
long-standing issue that just became obvious because of [1].
This CL does not attempt to fix the root cause but just tries to
suppresses the NPE to unblock QA tasks. A proper fix should be made in
subsequent CLs.
Bug: 22517687
Bug: 26511607
Change-Id: I6bc87c3d18b560fe2253fb9f05557b95b04d0cf0
When requested (-a or -S options), add to the meminfo dump the
proportional swap when available to the system. The reported amount is
the PSS memory not having being accessed recently and being swapped out
of memory.
Total PSS by process:
76,265K: system (pid 912) ( 290K in swap)
67,080K: com.android.systemui (pid 3133) ( 270K in swap)
...
Total PSS by category:
287,803K: .dex mmap ( 0K in swap)
264,575K: Native ( 25,184K in swap)
241,077K: Dalvik ( 9,391K in swap)
Bug: 26190445
Change-Id: I96415ec65b60c9f60e397d2b3acca9a6e1399aba
As seen in frameworks/support!
Previously we would not set a fragment's new state until the move to a
new target state was fully complete. This causes problems when other
parts of the fragment manager infrastructure (such as lazily
initializing a child fragment manager) read that state while we're
dispatching a state change call to a fragment.
In this situation, adding a child fragment and then calling
executePendingTransactions on the child FragmentManager would not have
the intended effect, as the child FragmentManager would still be in
state INITIALIZING. The expected lifecycle callbacks to the child
fragment would then occur later.
Fix this by updating the fragment state as we go through each phase of
moveToState before we dispatch to the associated onState method,
matching our usual pattern of invoking onFoo methods after foo has
occurred. Delete the redundant resumed field as we now can use the
state directly.
Bug 25019275
Change-Id: I97fe45578d59ab643c9779eaeb475a331e446335
willSaveLayer was removed in https://codereview.chromium.org/1537193002
Requires SkCanvas::SaveLayerFlagsToSaveFlags, added in ag/839104
(Ic78acd6d780245bd30b891e7dcabe29daacc17f4)
Change-Id: Idc49b38fc7939620267b4402ff3846b29f27c81a
(cherry picked from commit e208643a6f9d076274dd683c51a086b1e3679896)
Parse "SwapPss:" lines from /proc/pid/smaps if it exist, and store them
in a seperate stat entry.
Report SwapPss if made available by kernel, otherwise we fall back to
legacy Swap.
Fix getTotalSwappablePss documentation.
Change-Id: I361928c0f44c7dc9b959b91c127c916215063866
Signed-off-by: Thierry Strudel <tstrudel@google.com>
Uses consistent naming and ordering for padding and inset calculations.
Updates documentation to include UNDEFINED_INSET as public API, since we
return the value and will accept the value as a valid param. Flattens
out RTL logic for readability.
Change-Id: Ie704e2e5fc7f5763500ebe1217ddbbf4d0c5ef86
- New carrier config option is used to determine if presence is used
to determine whether a contact is capable of video calling.
- Also, improve logging for PhoneAccount capabilities.
Bug: 20257833
Change-Id: Ifcc7df95677eb4399f08eb8849c4004892957e90
The CL was previously reviewed at ag/842930.
> We can obtain detailed information of MtpEvent from devices. e.g. object
> handle of changed object. The CL adds the detailed information as
> properties of MtpEvent class.
>
> BUG=26480986
Change-Id: I93afad9caf118d74cd0923d70242133c4fb2a648
Can't safely rewind clip allocations, since those pointers are cached by
ClipArea. Instead add early rejection to handle most cases, and update
tests.
Change-Id: Ic32f95cf95602f427f25761a8da1583c4495f36d
