We used to guarantee that a layer in SurfaceFlinger would never be
destroyed before all references (to its ISurface) on the client
side would be released. At some point, this guarantee got
relaxed to allow to free gralloc resources sooner. This last
change was incorrect, because:
- in implementations with reference-counting the gralloc resources
wouldn't be released anyways, until all the mapping were gone
- in implementations without ref counting, the client side
would most likely crash or do something bad
- it also caused the SharedBufferStack slot to be reallocated
to another surface, which could be problematic if the client
continued to use the surface after the window manager destroyed it.
So, we essentially reinstate the guarantee that layers won't be
destroyed until after all references to their ISurface are
released.
NOTE: This doesn't entirely fix 3306150 because there is another
problem there where the Browser continues to use a surface after it
has been destroyed.
also improve SurfaceFlinger 'dumpsys' log
list the purgatory, which shows windows that have been closed,
but for which the client still has references.
we were not clearing the screen entirely, which caused garbage when
the screen wasn't entirely covered by windows.
Change-Id: Ia7aa13c36a8a314e0e8427d419b16b9aa2165ddf
we make sure to call compositionComplete after everytime we do
composition with the GPU (even for the screenshot case), which is
where the buffer locks are released.
Change-Id: I450430d1e4d1ee9ce1023970642378c42cdcfa4c
Also make sure they are non-negative.
This is not documented in Mifare Classic spec, but based on findings from NXP:
- Operand should be stored in little-endian format in the transceive buffer
- Tag ignores the sign bit on the operand, its effectively 31-bit unsigned
- Overflow and underflow generates an error.
Change-Id: Id3389b3894ded732c4b00d564ca53f5df651359e
Fixed a bug in the handling of SENSOR_LANDSCAPE and SENSOR_PORTAIT
on devices that have a natural landscape orientation. The old code
was disabling 180 degree orientation detection when it shouldn't have.
Bug: 3381359
Change-Id: I19bd2519e7f69a24835840fbf542aa7349981029
When generating a mock tag (after a NDEF exchange over LLCP), one of
the internal fields is set to null. This was causing NullPointerException
when being converted to a Parcel.
This is fixed by not including this field in the Parcel for mock tags.
Change-Id: I000e2faa54d71fd755ba7993e1e258743aad98fb
Remove method overloading for combinations of sector+block addressing.
Instead provide methods that more closly match the raw commands, and more
efficient helpers to convert between blocks and sectors.
o fix off-by-one bug in getBlockCountInSector()
o add BLOCK_SIZE
o remove DESFIRE not operating in classic emulation (SAK 0x20)
o hide isEmulated(), there is no use case, and the info is available elsewhere
o getTotalBlockCount() -> getBlockCount()
o getBlockCount(int) -> getBlockCountInSector(int)
o introduce blockToSector() and sectorToBlock()
o remove authenticateBlock()
make it really clear that authentication is per sector, and reduce function
explosion. blockToSector() allows you to use authenticateSector...
o explicit authenticateSectorWithKeyA() / authenticateSectorWithKeyB()
get rid of magic boolean
o remove all (int sector, int block) parameters
always address by absolute block. this makes the API crystal clear, and
helps reduce function explosion
o validation of all sector and block indices
o dont & 0xff when converting to byte - its redundant
o Remove TYPE_OTHER. Mifare Classic types are well-known and stable.
Change-Id: I3c9f8254ff307f31b388b3d7592c862d5de6afa5
This catches possible exceptions caused by attempting to
unregister a receiver that may have already been unregistered.
This is related to the OOM issue workaround in Change I99a7e0c3.
Change-Id: Ib69a8acc6b12659125c958a03efac0a8a68a3859
ACL auto connection was removed so other profiles
won't auto connect when priority was auto connect.
Reduced connect other profiles time delay.
Bug: 3281016
Change-Id: Ic56ca6e95f618f02bc316d1f80014e58c4b6c3bc
Wakelock will get released while
1) no request pending to be sent out, in which mRequestMessagesPending increases
before calling EVENT_SEND and decreases while handling EVENT_SEND.
and
2) no waiting requests sent to RIL but no replied, in which mRequestMessagesWaiting
increases while sending request and decreases while handling response.
Both will be cleared while WAKE_LOCK_TIMEOUT occurs to recovery from out of sync situation.
bug: 3369427, 3370827
Change-Id: Ib2fc54db3b155bd3fb1296ad83720b7836708caf
Bug: 3241808
Make sure the directory the caller is trying to open the icon db
in actually exists and create it if it doesn't.
Change-Id: Icc0a67b0ceba03a746ac2cd3b84fc34c2c8b4fe2
The creation of a TextToSpeech object causes the TTS service to
start, and the initialization listener to be called when the
service is connected. But the listener is never called when
the service binding failed to be notified of this error.
The fix consists in checking the result of the bind to service
operation, and notify the listener in case of an error.
More log was added in case speak() and synthesizeToFile() are
called but the service is not known to have started.
Change-Id: I7dcc1fa44be31fee3177ec6215fca3306377b934
EffectModule::process() was copying effect chain input buffer to
output buffer if no effect was active instead of accumulating it.
Change-Id: If4ca75601ea69a088d0f71d88aec53e90a1dec89
The latter is an operation that otherwise could block on the datasource
for a significant amount of time. During that time we'd be unable to
abort the preparation phase without this prefill.
Change-Id: I3bc889b264f599bfd5c2bbdf48b88ccb55d86172
related-to-bug: 3362836
The cut-off frequency of the lowpass filter was too high
for the sampling rate used by DELAY_NORMAL.
Now we use the same filters used for the gravity vector
(cascaded biquad at 1.5 Hz)
Change-Id: Iac290a716cc47a78337a8f0e45b103e49b4d9d78
Fixes a regression from Froyo. Previously, when a scroll view
had no children, its onTouchEvent would return because the scroll
view would only start dragging if the user touched one of its children.
In Gingerbread, the user can drag from anywhere within the scroll
view, not just by touching a child. However, it makes no sense
to drag a scroll view that has no children so an empty scroll view
should just ignore touches like any other empty view group would.
This change fixes applications that for some reason or other happen
to have empty scroll views in strange places.
Bug: 3246230
Change-Id: I76c6136d3cd74968983033014d60deec1718ce30
