The warnings in these files were hidden by -isystem
framework/native/include.
Bug: 31752268
Test: m -j
Change-Id: I2a54376aea380ee24e6483fb7d35fdfe8991c490
system/core/include is included in the global include path using
-isystem, which hides all warnings. Fix warnings in libutils
headers in preparation for moving from -isystem to -I.
- Fix implicit cast from int64_t to long in Condition.h. Remove
the __LP64__ check and always compare against LONG_MAX before
casting.
- Fix implicit cast from size_t to ssize_t in KeyedVector.h
- Fix -Wshadow-field-in-constructor warnings in Looper.h and RefBase.h
- Move destructors for MessageHandler and LooperCallback to Looper.cpp
and ReferenceRenamer and VirtualLightRefBase to RefBase.cpp to prevent
vtables in every compilation unit.
- Declare template variables in Singleton.h
- Fix old-style casts in StrongPointer.h and TypeHelpers.h
- Use template metaprogramming in TypeHelpers.h to avoid warnings on
memmove on non-trivial types.
- Add an assignment operator to key_value_pair_t to complete
rule-of-three
- Use memcpy instead of dereferencing a reinterpret_casted pointer to
treat the bits of a float or double as int32_t or int64_t
- Escape unicode sequences inside doxygen comments between \code and
\endcode
- Remove WIN32 ZD definition in Compat.h, %zd works fine with mingw
- Fix WIN32 printf warnings in Filemap.cpp
- Initialize mNullValue with 0 in LruCache.h, some of the tests use a
non-pointer type for TValue.
Test: m -j native
Bug: 31492149
Change-Id: I385a05a3ca01258e44fe3b37ef77e4aaff547b26
This prevents two different kinds of client errors from causing
undetected memory corruption, and helps with the detection of others:
1. We no longer deallocate objects when the weak count goes to zero
and there have been no strong references. This otherwise causes
us to return a garbage object from a constructor if the constructor
allocates and deallocates a weak pointer to this. And we do know
that clients allocate such weak pointers in constructors and their
lifetime is hard to trace.
2. We abort if a RefBase object is explicitly destroyed while
the weak count is nonzero. Otherwise a subsequent decrement
would cause a write to potentially reallocated memory.
3. We check counter values returned by atomic decrements for
plausibility, and fail immediately if they are not plausible.
We unconditionally log any cases in which 1 changes behavior
from before. We abort in cases in which 2 changes behavior, since
those reflect clear bugs.
In case 1, a log message now indicates a possible leak. We have
not seen such a message in practice.
The third point introduces a small amount of overhead into the
reference count decrement path. But this should be negligible
compared to the actual decrement cost.
Add a test for promote/attemptIncStrong that tries to check for
both (1) above and concurrent operation of attemptIncStrong.
Add some additional warnings and explanations to the RefBase
documentation.
Bug: 30503444
Bug: 30292291
Bug: 30292538
Change-Id: Ida92b9a2e247f543a948a75d221fbc0038dea66c
Add basic interface documentation to RefBase.h.
Much, but not all, of this is cut-and-pasted from an email message
from Mathias Agopian. The rest is reconstructed from the code.
Delete some, now redundant, text from Refbase.cpp, and add a bit
more about the implementation strategy.
Some minor fixes to internal comments.
Bug: 30292291
Change-Id: I56518ae5553bc6de0cc2331778e7fcf2e6c4fd87
We comment out the names of the parameters we're intentionally
not using.
(cherrypick of 44a0eb4d8cdf4ab881f5d5a1acc1155266a46f6a.)
Change-Id: I0a92d9200ca26cdc4700d2ff9e3ab33c102520d5
Convert to use std::atomic directly.
Consistently use relaxed ordering for increments, release ordering
for decrements, and an added acquire fence when the count goes to
zero.
Fix what looks like another race in attemptIncStrong:
It seems entirely possible that the final adjustment for
INITIAL_STRONG_VALUE would see e.g. INITIAL_STRONG_VALUE + 1,
since we could be running in the middle of another initial
increment.
Attempt to somewhat document what this actually does, and
what's expected from the client. Hide the documentation in
the .cpp file for now.
Remove a confusing redundant test in decWeak. OBJECT_LIFETIME_STRONG
and OBJECT_LIFETIME_WEAK are the only options, in spite of some
of the original comments.
It's conceivable that either of these issues has resulted in
actual crashes, though I would guess the probability is small.
It's hard enough to reason about this code without the bugs.
Bug: 28705989
Change-Id: I4107a56c3fc0fdb7ee17fc8a8f0dd7fb128af9d8
This patch makes the Renamer class destructor virtual and fixes
the incorrect constructor initialization list ordering.
These issues and related compiler warnings have been suppressed
by the build system which uses the gcc isystem command
line option, unless a project explicitly adds the include the path
system/core/include to Android.mk and uses the templates.
Change-Id: Iff76a655eb8bd547adfe994c7315a005e98aed41
Signed-off-by: Ukri Niemimuukko <ukri.niemimuukko@intel.com>
Signed-off-by: Mingwei Shi <mingwei.shi@intel.com>
Signed-off-by: Yong Yao <yong.yao@intel.com>
background:
we have some code to fix-up the IDs of references when
using RefBase's DEBUG_REFS when those refs are managed by
arrays wp<> or sp<> (this is because wp<> / sp<> don't have
a trivial ctor when DEBUG_REFS is enabled, and Vector
treats them as trivial for obvious performance reasons)
this is complicated by the fact that we don't want to have
to recompile everything when enabling DEBUG_REFs (i.e.: the
Vector code cannot know wheter it's enabled or not for its
template stuff).
problem:
there was a bug in the fix-up code for wp<> which was trying
to access the weakref_impl from the RefBase* however, this was
moronic since RefBase could have been destroyed if there wasn't
any more strong refs -- and this happned. Instead we need to get
the weakref_impl directly from the wp<>
Change-Id: Ie16e334204205fdbff142acb9faff8479a78450b
Many of our basic data structures are trivially movable using
memcpy() even if they are not trivially constructable, destructable
or copyable. It's worth taking advantage of this *ahem* trait.
Adding trivial_move_trait to String16 reduces appt running
time on frameworks/base/core/res by 40%!
Change-Id: I630a1a027e2d0ded96856e4ca042ea82906289fe
we would leak a weakref_impl if a RefBase was never incWeak()'ed.
there was also a dangling pointer that would cause memory corruption
and double-delete when a custom destroyer was used to delay the
execution of ~RefBase.
it turns out that the custom destroyer feature caused most of the
problems, so it's now gone. The only client was SurfaceFlinger
who now handles things on its own.
RefBase is essentially back its "gingerbread" state, but the
code was slightly cleaned-up.
Bug: 5151207, 5084978
Change-Id: Id6ef1d707f96d96366f75068f77b30e0ce2722a5
Also revert all dependent changes:
This reverts commit 8e18668d14adf601cbe5973030c310ec23d88461.
This reverts commit 69b4587bfbb3e98f793959d9123340360fa233a2.
This reverts commit a9c9a4baf24700e8817d47d8ea8da1742caea0b5.
This reverts commit 2c0042b666a969091c931614f2fc0dce2f1cfac8.
This reverts commit f6c8206735e7e078461e5f2aef6e1a1446fdd075.
This reverts commit 24855c09173a6caaec7dcedd0c2d7ce15121d39b.
Change-Id: I33e699640f3f59e42fa03c99a9a1b7af0d27d4d8
This adds a destroy() virtual on RefBase which
sublasses can implement. destroy() is called
in lieu of the destructor whenthe last strong
ref goes away.
this was introduced recently. we make sure to use
the correct owner id (the sp) instead of the wp.
Change-Id: I78fdc6ec0c2d3e687278b70442d74d1924b512a2
First slipt sp<> out of RefBase into StrongPointer.h so it can be reused
more easily and to make it clear that it doesn't require RefBase.
Note: the rest of the change only affects the system when DEBUG_REFS is enabled.
The main problem we fix here is that the owner id associated with each
reference could get out of date when a sp<> or wp<> was moved, for
instance when they're used in a Vector< >.
We fix this issue by calling into RefBase::moveReferences from
a template specialization for sp<TYPE> and wp<TYPE> of the
type helpers. RefBase::moveReferences() has then a chance to
update the owner ids.
There is a little bit of trickery to implement this generically in
RefBase, where we need to use a templatized functor that can turn
a sp<TYPE>* casted to a void* into a RefBase*.
Introduced a new debug option DEBUG_REFS_FATAL_SANITY_CHECKS
currently set to 0 by default as there seem to be an issue
with sp<ANativeWindow> which trips the sanity checks.
Change-Id: I4825b21c8ec47d4a0ef35d760760ae0c9cdfbd7f
weak pointer comparison operators were implemented wrong,
they were using the internal "unsafe" pointer. We could end up
with two "equal" weak pointer pointing to different objects.
this caused KeyedVector keyed by weak pointer to work incorrectly, in
turn causing a window that just got added to a list to be immediately
removed.
Change-Id: Ib191010c39aafa9229109e4211f6c3b2b2f9696d
when assigning a smart pointer to another one, we need to make
sure to read all the data we need from the right-hand-side
reference (the assignee) before we decRef the assigned.
This bug would cause linked-list of smart-pointers to fail
miserably.
Change-Id: Ibb554c15fddf909f7737c632b7c80322e80ea93f
- Currently the lock/unlock path is naive and is done for each drawing operation (glDrawElements and glDrawArrays). this should be improved eventually.
- factor all the lock/unlock code in SurfaceBuffer.
- fixed "showupdate" so it works even when we don't have preserving eglSwapBuffers().
- improved the situation with the dirty-region and fixed a problem that caused GL apps to not update.
- make use of LightRefBase() where needed, instead of duplicating its implementation
- add LightRefBase::getStrongCount()
- renamed EGLNativeWindowSurface.cpp to FramebufferNativeWindow.cpp
- disabled copybits test, since it clashes with the new gralloc api
- Camera/Video will be fixed later when we rework the overlay apis
Colin Cross