| bbbb6e80422eba9770be7410ba2521e0dfe35606 |
|
03-Jun-2014 |
Jeff Hao <jeffhao@google.com> |
Workaround to tolerate void methods with non-zero parameters_off. - DO NOT MERGE
Bug: 15259852
Change-Id: Ibf5903b6a1c88fa6df4e5ef9c01351ab899df1f8
|
| c5b2915ad8865cc1907217e4b3dfdb838c3c843d |
|
31-Mar-2014 |
Neil Fuller <nfuller@google.com> |
Fix a copy and paste doc error
Change-Id: I2a142f0a6e8348993b37666876d4f812a7e69986
|
| 08da684abe6d5515a9c760a3a33af4f8b426095d |
|
26-Mar-2014 |
Neil Fuller <nfuller@google.com> |
Improve the performance of Annotation checks
In the circumstance where the annotated element (field/
method/constructor/class) was loaded from a different Dex file
than the annotation:
The previous code would perform a binary
search to determine the type ID of the annotation in the Dex
file of the annotated element. That type ID would then be
compared against the type IDs of annotations present on the
annotated element. The binary search was quite expensive: it
involved various indirections, many small native method calls
and temporary String creation.
Instead, the code now compares the names of the
annotations on the annotated element with the name of the
annotation being searched for. The name of the annotation
is known and cached on the class and is therefore cheap to
get. The name was previously being used to binary search for
the type ID so this appears to be no less correct.
Also removed some unused methods in order to delete the
getFieldIndex() method.
Change-Id: Ib8fb021ddf1221e3eac983aa87e7aea8174720ef
|
| a6e22fc9b70ebe39abd716ce37450bda935c0fb8 |
|
19-Sep-2013 |
Ian Rogers <irogers@google.com> |
Avoid computing class def indices.
Bug: 10244719
Also tidy AnnotationAccess.
(cherry-picked from 8b2c0b9abc3f520495f4387ea040132ba85cae69)
Change-Id: I6ec8fd4e36b428d7e16e01d98b7bebc143fac8c3
Conflicts:
libdvm/src/main/java/java/lang/Class.java
|
| d4ec55c4e1acc3c3df937facbd367aff6618536c |
|
07-Sep-2013 |
Ian Rogers <irogers@google.com> |
Lazier annotation signature parsing.
Bug: 10244719.
Also, make wider use of empty arrays.
Also, use named inner classes to aid profiling.
Remove caching as the empty result will be fast.
Porting the ART changes to DVM, to avoid the caching, results in bringing
in a bunch of other ART clean-up.
TODO, avoid modified-UTF8 to String conversions by Dex.
Change-Id: I051b6e34b5a4e7a7f1383461a799755947a5ba36
Conflicts:
libdvm/src/main/java/java/lang/Class.java
|
| d3e311496e0b08acd959c3ffa31d9930a71bae19 |
|
24-Jul-2013 |
Ian Rogers <irogers@google.com> |
Improve the performance of reflective operations.
Common dex index lookups require a number of different types to be constructed,
implement short-cuts that avoid the intermediate data types.
Refactor reflective operations to use dex methods that don't require temporary
objects.
Note: this change adds an empty short array to Dex, this code can't appear in
libcore.util as Dex doesn't always come with libcore.util.
Bug: 9968234
(cherry picked from commit 381db8de0ac5df124c7f482bf4be9d5756bf3ca4)
Change-Id: I9122a3f01b4f41b07a84aff768193c092e29698c
|
| 00a2d1ee4cfee1f33344c3798940bbae22f96187 |
|
05-Sep-2012 |
Benoit Lamarche <benoitlamarche@google.com> |
Fix an ugly bug where try/catch offsets weren't being mapped properly
This CL comes from CL I3bdadf20899fdb5d4d074e69103b33c0404a31f8 in AOSP master.
Modifications have been made to files that were in dalvik but are now in libcore.
Here is the original commit message:
In dex files, there are several places where one object refers
to another object by its position in the file. The dex merger
is generally very careful to adjust these mappings when combining
dex files.
Unfortunately one of these cases was broken. Each try_item refers
to a corresponding encoded_catch_handler by its byte offset in a
list. Most of the time this byte offset is the same in the input
dex file and the output dex file. But encoded_catch_handlers are
variable-length because they use a variable-length uleb128 encoding
to address the type_idx being caught. When dex files are merged,
some exception types may go from having a small index to having
a large index, increasing the number of bytes required to encode
that index. This breaks our ability to directly copy over offsets
as we were doing previously.
Change-Id: I61adb8fce6fa9b83cbfbe5eef7edea84664d7ea6
(cherry picked from commit b44450d600807ab130aa7c6d91b534fde52e0a21)
|
| dc5a528c10462726682b8f490772e0677b038c1f |
|
08-Jun-2012 |
Elliott Hughes <enh@google.com> |
*libcore* resolved conflicts for merge of bc23c4f3 to dalvik-dev
DexBuffer became Dex, and moved to libcore.
Change-Id: I32447f83c03dd18f2339e879d015134be9f4fc18
(cherry picked from commit 621d004d61b3eaca74fc045a3fba6afa14327a2e)
|
| 0436f436f122e7e74285faf32b8db2259f56ded7 |
|
28-Mar-2012 |
Jesse Wilson <jessewilson@google.com> |
Manually merge "Compact merged dexes on their byte aligned size"
The changed file was moved from dalvik/ to libcore/ in this branch.
See https://android-review.googlesource.com/34010
Change-Id: I9b2fae8767fb4e7461e9b762e70efb4171e7a9f5
(cherry picked from commit 20d13d20f6b8d4094d4d66bb692ac19c4b81348f)
|
| a34f79ba9d4983b96571bbde0a18d352d495f601 |
|
06-Oct-2011 |
Jesse Wilson <jessewilson@google.com> |
Make Dex access faster by making binarySearch work.
Change-Id: I5f506ef2cd73c2f88f395ad1e1200d79ed33f654
(cherry picked from commit 8cf91498080408179f1b96ae832b581b8547d7c4)
|
| 450a197f09d35a07be12a924d4db1ecd2bee65fe |
|
30-Apr-2013 |
Brian Carlstrom <bdc@google.com> |
Remove redundant alignToFourBytes in favor of alignToFourBytesWithZeroFill
(cherry picked from commit d2301a66b2629f2841ab3a3cc7684a645ebe7217)
Change-Id: I32522d86950d0c1a89e77b44ddeab071ce9c021a
|
| df60589435658db76be61708bfa66515ffba40af |
|
30-Apr-2013 |
Brian Carlstrom <bdc@google.com> |
Fix Dex.create(ByteBuffer) to not allocate a byte[] to check magic
(cherry picked from commit 34c119762d55411ec2451b38f4782b923217aa2c)
Change-Id: I968762e1962fb720bbcf0c2866f17cd0500a9e54
|
| 030bd62e38ba41b8f464f009382955c51bd72f04 |
|
19-Sep-2011 |
Jesse Wilson <jessewilson@google.com> |
Switch Dex from byte[] to ByteBuffer for backing store.
The ByteBuffer allows Dex to share the mapped data with
Class.getDex().
This also changes Dex.Section to use a ByteBuffer; that
ByteBuffer's position and limit are used instead of a
separate pair of integers.
Change-Id: I2b8042a493f70213c30e5c26e0bacfb03daebf55
(cherry picked from commit 59653899ed7cbca17f5a3dd32a23e81b3c4a5339)
|
| c69e85bfb7ddd6c4b0023ca29eeb0416087f9a8b |
|
30-Apr-2013 |
Brian Carlstrom <bdc@google.com> |
Add Dex.create(ByteBuffer)
(cherry picked from commit aca7a4ba7597be831164d94d787f383b8d4ac2fd)
Change-Id: Id517f2ccda31a20e28fe63005120e4e975935980
|
| 2bea5ee615b0f4add658d5660bd81c5145a0d05e |
|
18-Sep-2011 |
Jesse Wilson <jessewilson@google.com> |
Move dex utilities from dalvik/dx to libcore/dex part 2
Change-Id: Ib5bb7bec80d77464f632f1cdfef708d868301c42
(cherry picked from commit 9ae95d8f3381204108abc722e8e5a03d6ecbf311)
|