GitLab

Bug: 5934554
Change-Id: I640cd54c227df2bf662d484cb2af95ece4d13421

We were throwing under the premise that we'd always have
response headers before we had a cached body. But when the
user calls connect() we'll have queried the cache but we
won't necessarily have response headers from the server yet.
That's because connect() doesn't require a response to be ready.

Bug: http://b/5920659
Cherry-picked from: I9d69b44b5be3d0e29f14d8b83781f6e65c67582a

Change-Id: Ia403b42ddc9783fd1456164f4b7350da3ce80566

Stores 0x0 and Float.MIN_VALUE (0x1) into the same register and makes
sure it can still be used as float.

(cherry picked from commit ffa7f346)

Change-Id: Ia72f24f1233f7c0a8a794def8cbe94151b60b2af

java.net.Socket creates the socket on connect(). But in case of failure
during the connect() it closes the socket.
During the create, BlockGuard's socket() is the one tagging the socket.
But in case of failure, nobody untags that socket.
So now we untag scokets on close() in BlockGuard

Bug: 5274621
Change-Id: I282665a05e2dc184df77c84ceab49fb55b7695af

Bug 5498103

Use a StreamHandler instead of a FileHandler, so that FileHandler
doesn't throw an exception if the log directory doesn't exist.

Change-Id: I7b7334ff8290a4b028712e89bb432fd78ef48346

Bug: 5466273

(cherry picked from commit d3433cea)

Change-Id: I8618be21a2227d66ea66352342b530906605160f

Bug: 5274621
Change-Id: Icc42a80c8cce6adb20d16ca2bb1fb51e2a8e8e1b

The kernel uses awkward device-specific units that make tests with timeouts
that aren't integer multiples of a second awkward.

There's no common superclass for the various sockets, so we can't factor
out the test even though -- apart from the creation of the specific socket --
it's textually identical in each case. Nice library design, boys!

I've also fixed a couple of CloseGuard warnings in passing.

Bug: 5159133
Change-Id: Iaf4c9891b3cf2f5b3ebcee5cb2ba862157ab034e

The kernel uses awkward device-specific units that make tests with timeouts
that aren't integer multiples of a second awkward.

There's no common superclass for the various sockets, so we can't factor
out the test even though -- apart from the creation of the specific socket --
it's textually identical in each case. Nice library design, boys!

I've also fixed a couple of CloseGuard warnings in passing.

Bug: 5159133
Change-Id: Iaf4c9891b3cf2f5b3ebcee5cb2ba862157ab034e

Whatever the root cause of the fstat(2) failure, we should still attempt the
requested close(2) --- even if it's just going to fail in the same way the
fstat(2) did.

Bug: 5407056
Change-Id: Ie988d1b1ebdb6b378fb537b2f56245fab568ee7c

* commit '64350311':
  Work around a dalvik weakness.

Our register maps aren't good enough, so when the collector looks at the
FinalizerReference for the RandomAccessFile, the referent has been marked
because the RandomAccessFile is still pointed to by a dex register, and
the register maps don't take into account that that register can never be
accessed again after the current dex PC, so it should in fact not be marked.

This is a long-standing weakness (I can cause similar failures in Gingerbread).

Bug: 5412580
Change-Id: I1de253a3686701ee724f8fefd5a6230d11179bad

* commit '9abfeaca':
  Fix backwards Engine cache test

Also fix SignatureTestMD2withRSA to clean up its test provider.
Add context information on failures to CertificateTest.

Bug: 5479141
Change-Id: I9a5ce60cfda63cfc418e72464eb1d3ef81dab0c0

am bea6b3d2: Merge "Don\'t require exception priority to match any particular implementation." into ics-mr0

* commit 'bea6b3d2':
  Don't require exception priority to match any particular implementation.

Bug: http://b/5478266
Change-Id: I15ad1f1eb67b307b61819c0c2031cc9a5f2e00fd

Change-Id: If10cf39515472969a2e57e9a1e00dead818fe771

* commit '0aaf7fc3':
  Set ANDROID_DATA for DalvikExecTest

* commit '5576e932':
  Fix Arrays Javadoc to be consistent with parameter names.

* commit '5de85581':
  DexClassLoaderTest.setUp should clean optimized directory