| e5de54cfab5f14ba0b8ff25d8d60901c7021943f |
|
20-Apr-2016 |
Calin Juravle <calin@google.com> |
Split profile recording from jit compilation
We still use ProfileInfo objects to record profile information. That
gives us the flexibility to add the inline caches in the future and the
convenience of the already implemented GC.
If UseJIT is false and SaveProfilingInfo true, we will only record the
ProfileInfo and never launch compilation tasks.
Bug: 27916886
Change-Id: I6e4768dc5d58f2f85f947b276b4244aa11ce3fca
|
| 09ed09866da6d8c7448ef297c148bfa577a247c2 |
|
12-Feb-2016 |
David Srbecky <dsrbecky@google.com> |
Pack stack map entries on bit level to save space.
Use only the minimum number of bits required to store stack map data.
For example, if native_pc needs 5 bits and dex_pc needs 3 bits, they
will share the first byte of the stack map entry.
The header is changed to store bit offsets of the fields rather than
byte sizes. Offsets also make it easier to access later fields without
calculating sum of all previous sizes.
All of the header fields are byte sized or encoded as ULEB128 instead
of the previous fixed size encoding. This shrinks it by about half.
It saves 3.6 MB from non-debuggable boot.oat (AOSP).
It saves 3.1 MB from debuggable boot.oat (AOSP).
It saves 2.8 MB (of 99.4 MB) from /system/framework/arm/ (GOOG).
It saves 1.0 MB (of 27.8 MB) from /system/framework/oat/arm/ (GOOG).
Field loads from stackmaps seem to get around 10% faster.
(based on the time it takes to load all stackmap entries from boot.oat)
Bug: 27640410
Change-Id: I8bf0996b4eb24300c1b0dfc6e9d99fe85d04a1b7
|
| bf9611f821697b14bf9e170f503c3f47613b046b |
|
26-Mar-2016 |
Andreas Gampe <agampe@google.com> |
ART: Clean up verifier
Clean up verifier post-Quick.
Change-Id: I0b05e10dd06edd228fe2068c8afffc4b7d7fdffa
|
| 7dc11782ff0a5dffcd8108f256f8975f0b3e8076 |
|
25-Feb-2016 |
David Srbecky <dsrbecky@google.com> |
Implement << operator for DexRegisterLocation::Kind.
This makes it comparable in DCHECK_EQ and similar methods.
Change-Id: I6b5b237be89325850ae6860d011fd6741189ab01
|
| b52de2427700ebdecc904aa2f3f59a079319abd5 |
|
19-Feb-2016 |
Nicolas Geoffray <ngeoffray@google.com> |
Fix braino in single frame deopt.
handler_method_header_ is actually the OatQuickMethodHeader
of the caller of the method that is deoptimizing (as we will
resume at its call site). We should use the OatQuickMethodHeader
of the method that is deoptimizing and pass it to the code cache
to invalidate the method.
Change-Id: I2b61a96b8415e9477e3ac902c54a314aeb0811a0
|
| b88d59ef4fe611fe47e50a6a19785e03bbd5f93b |
|
17-Feb-2016 |
Nicolas Geoffray <ngeoffray@google.com> |
Be a bit smarter with JIT code triggering deoptimization.
Do not re-use an OSR method that triggered deoptimization.
Also add a stack overflow check before doing OSR.
bug:27094810
Change-Id: I6ff6a7fb9b3df9b7c0ff37e3610595efa70ad067
|
| 012fc4e9d9b66b3ffb7838b0e29dadbb4863ee69 |
|
08-Jan-2016 |
Nicolas Geoffray <ngeoffray@google.com> |
Don't encode a DexRegisterMap if there is no live register.
Change-Id: I76a291e6a0ac37f0590d16c7f5b866115588bc55
|
| 73be1e8f8609708f6624bb297c9628de44fd8b6f |
|
17-Sep-2015 |
Nicolas Geoffray <ngeoffray@google.com> |
Inline monomorphic calls.
Change-Id: If38171c2dc7d4a4378df5d050afc4fff4499c98f
|
| efc3f02504e8e3768cc18ab5347c5a6d19d6a935 |
|
28-Oct-2015 |
David Brazdil <dbrazdil@google.com> |
ART: Fix deopt from optimized code under inlining
Deoptimization of inlined frames would use the outer method's
vreg map, thus starting interpreter in bogus state.
Bug: 25331616
Change-Id: I1d4aefc731bb1386f3e6186a89b59981836480b3
|
| 338565069cc221cb73ba55dc586c8ab8dc0d592c |
|
20-Oct-2015 |
Nicolas Geoffray <ngeoffray@google.com> |
Do not run the verifier when deopting optimized frames.
The stack maps contain all the required information.
Change-Id: I709e72b3de0d76a4203a3a004f093071e7865642
|
| 524e7ea8cd17bad17bd9f3e0ccbb19ad0d4d9c02 |
|
16-Oct-2015 |
Nicolas Geoffray <ngeoffray@google.com> |
Remove ArtCode.
- Instead use OatQuickMethodHeader.
- Various cleanups now that we don't have all those
ArtMethod -> ArtCode -> OatQuickMethodHeader indirections.
As a consequence of this cleanup, exception handling got a bit
faster.
ParserCombinators benchmark (exception intensive) on x64: (lower is better)
Before:
ParserCombinators(RunTime): 1062500.0 us.
After:
ParserCombinators(RunTime): 833000.0 us.
Change-Id: Idac917b6f1b0dc254ad68fb3781cd61bccadb0f3
|
| 6bc4374e3fa00e3ee5e832e1761c43e0b8a71558 |
|
12-Oct-2015 |
Nicolas Geoffray <ngeoffray@google.com> |
Add an abstraction over a compiled code.
That's just step 1, moving code-related functions of ArtMethod to
another class. That class is only a wrapper on an ArtMethod, but will
be changed to be a wrapper around compiled code.
Change-Id: I6f35fc06d37220558dff61691e51ae20066b0dd6
|
| 639bdd13993644a267f177f8f5936496bda65e2b |
|
03-Jun-2015 |
Andreas Gampe <agampe@google.com> |
ART: Single-frame deopt
Add deoptimization of a single frame. Works by removing the managed
code frame and jumping into the quick-to-interpreter bridge, and
the bridge understanding a stored ShadowFrame.
We need a separate fixup pass. For x86, we leave the return address
on the stack so we don't need to push it there.
Bug: 21611912
Change-Id: I06625685ced8b054244f8685ab50b238a705b9d2
|
| 99170c636dfae4908b102347cfe9f92bad1881cc |
|
06-Jul-2015 |
Mingyao Yang <mingyao@google.com> |
Deoptimization support in optimizing compiler for setting local values
Due to compiler optimizations, we may not always be able to update
the value of a local variable in a compiled frame (like a variable
seen as constant by the compiler). To avoid that situation, we simply
deoptimize compiled frames updated by the debugger so they are
executed by the interpreter with the updated value.
When the debugger attempts to set a local variable (actually a DEX
register or a pair of registers) in a compiled frame, we allocate a
ShadowFrame associated to that frame (using its frame id) and set the
new value in that ShadowFrame. When we know we are about to continue
the execution of the compiled frame, we deoptimize the stack using
the preallocated ShadowFrame (instead of creating a new one). We
initialize it with the current value of all DEX registers except
the ones that have been set by the debugger. Therefore, the
ShadowFrame represent the runtime context modified by the debugger.
Bumps oat version to force recompilation.
Bug: 19944235
Change-Id: I0ebe6241264f7a3be0f14ee4516c1f7436e04da6
|
| 72f7b880d5d0057b9fac3a51ef3a0f22909bc633 |
|
15-Sep-2015 |
David Brazdil <dbrazdil@google.com> |
ART: Fix 004-ReferenceMap run test
This patch adds a new option to ArtMethod::ToNativeQuickPc to select
the order of iteration over stack maps. The method is only used by
the runtime to find native_pc of catch blocks, but also by the
004-ReferenceMap test which uses it to find native_pc of a safepoint.
Change-Id: Idb2b34aabf1ac7249c30a00806af7d63d7e682dd
|
| 77a48ae01bbc5b05ca009cf09e2fcb53e4c8ff23 |
|
15-Sep-2015 |
David Brazdil <dbrazdil@google.com> |
Revert "Revert "ART: Register allocation and runtime support for try/catch""
The original CL triggered b/24084144 which has been fixed
by Ib72e12a018437c404e82f7ad414554c66a4c6f8c.
This reverts commit 659562aaf133c41b8d90ec9216c07646f0f14362.
Change-Id: Id8980436172457d0fcb276349c4405f7c4110a55
|
| 26f728661a08062a373a3203b72dc2555c2aed2d |
|
15-Sep-2015 |
Sebastien Hertz <shertz@google.com> |
Cleanup thread access in StackVisitor
Adds method StackVisitor::GetThread to give access to the visited
Thread* so we no longer need to copy that pointer in subclasses.
Also adds a few missing const and DISALLOW_COPY_AND_ASSIGN.
Change-Id: I57649ee7742ef4ef1e01447ac2fbb66f977b22eb
|
| 659562aaf133c41b8d90ec9216c07646f0f14362 |
|
14-Sep-2015 |
David Brazdil <dbrazdil@google.com> |
Revert "ART: Register allocation and runtime support for try/catch"
Breaks libcore test org.apache.harmony.security.tests.java.security.KeyStorePrivateKeyEntryTest#testGetCertificateChain. Need to investigate.
This reverts commit b022fa1300e6d78639b3b910af0cf85c43df44bb.
Change-Id: Ib24d3a80064d963d273e557a93469c95f37b1f6f
|
| b022fa1300e6d78639b3b910af0cf85c43df44bb |
|
20-Aug-2015 |
David Brazdil <dbrazdil@google.com> |
ART: Register allocation and runtime support for try/catch
This patch completes a series of CLs that add support for try/catch
in the Optimizing compiler. With it, Optimizing can compile all
methods containing try/catch, provided they don't contain catch loops.
Future work will focus on improving performance of the generated code.
SsaLivenessAnalysis was updated to propagate liveness information of
instructions live at catch blocks, and to keep location information on
instructions which may be caught by catch phis.
RegisterAllocator was extended to spill values used after catch, and
to allocate spill slots for catch phis. Catch phis generated for the
same vreg share a spill slot as the raw value must be the same.
Location builders and slow paths were updated to reflect the fact that
throwing an exception may not lead to escaping the method.
Instruction code generators are forbidden from using of implicit null
checks in try blocks as live registers need to be saved before handing
over to the runtime.
CodeGenerator emits a stack map for each catch block, storing locations
of catch phis. CodeInfo and StackMapStream recognize this new type of
stack map and store them separate from other stack maps to avoid dex_pc
conflicts.
After having found the target catch block to deliver an exception to,
QuickExceptionHandler looks up the dex register maps at the throwing
instruction and the catch block and copies the values over to their
respective locations.
The runtime-support approach was selected because it allows for the
best performance in the normal control-flow path, since no propagation
of catch phi values is necessary until the exception is thrown. In
addition, it also greatly simplifies the register allocation phase.
ConstantHoisting was removed from LICMTest because it instantiated
(now abstract) HConstant and was bogus anyway (constants are always in
the entry block).
Change-Id: Ie31038ad8e3ee0c13a5bbbbaf5f0b3e532310e4e
|
| 520633bebd2bf4d70884d30f179dbde9f275aac6 |
|
08-Sep-2015 |
Sebastien Hertz <shertz@google.com> |
Support deoptimization on exception
Allows to deoptimize when an exception is being thrown. We only
deoptimize if an executable frame (starting from the catch handler)
needs to be executed with the interpreter.
Before executing deoptimized frames, the exception is restored. The
interpreter starts by handling this exception at the point of the
throwing instruction.
Bug: 23714835
Change-Id: I0c5f7d4b257644acf12210aae8e5b6bb0f4af1f7
|
| b995c0ab7d6313f32e704e79c1e7dd21f5f160d4 |
|
24-Aug-2015 |
Sebastien Hertz <shertz@google.com> |
Update instrumentation stack after exception event
If instrumentation is updated on an exception event, we must update
the instrumentation stack (to remove all frames that will be unwound
by the exception) before doing the long jump.
Bug: 23471864
Change-Id: Ic91552bb7280c54bcc58b7ba03a17040b0b0f5ef
|
| 90443477f9a0061581c420775ce3b7eeae7468bc |
|
17-Jul-2015 |
Mathieu Chartier <mathieuc@google.com> |
Move to newer clang annotations
Also enable -Wthread-safety-negative.
Changes:
Switch to capabilities and negative capabilities.
Future work:
Use capabilities to implement uninterruptible annotations to work
with AssertNoThreadSuspension.
Bug: 20072211
Change-Id: I42fcbe0300d98a831c89d1eff3ecd5a7e99ebf33
|
| bf1fa2ccb5e7409910b99dc46b616e44c66ade68 |
|
09-Jun-2015 |
Sebastien Hertz <shertz@google.com> |
Follow up on CL 151605
- Fixes return type of StackedShadowFrameRecord::GetType
- Makes StackedShadowFrameType an enum class (scoped enum)
- Moves DeoptimizationReturnValueRecord and StackedShadowFrameRecord
to thread.cc file and use forward declaration in thread.h header
- Fixes tools/generate-operator-out.py for scoped enum classes.
Bug: 20845490
(cherry picked from commit f795869da0a1fa006fdcdacd8afb6149a63fc1a7)
Change-Id: I6b67e288b1db563699161e58ec2e2330d42dd8f5
|
| ef484d442a3dcae2cd1842c5be0623f5cf71e4ab |
|
18-May-2015 |
Mingyao Yang <mingyao@google.com> |
Fix nested deoptimization.
Handle nested deoptimization cases. Create a stacked shadow frame
records to keep track of deoptimization shadow frames. Shadow frames
under construction can be tracked in the same stack.
Bug: 20845490
(cherry picked from commit 1f2d3ba6af52cf6f566deb38b7e07735c9a08fb6)
Change-Id: I768285792c29e7c3cfcd21e7a2600802506024d8
|
| f795869da0a1fa006fdcdacd8afb6149a63fc1a7 |
|
09-Jun-2015 |
Sebastien Hertz <shertz@google.com> |
Follow up on CL 151605
- Fixes return type of StackedShadowFrameRecord::GetType
- Makes StackedShadowFrameType an enum class (scoped enum)
- Moves DeoptimizationReturnValueRecord and StackedShadowFrameRecord
to thread.cc file and use forward declaration in thread.h header
- Fixes tools/generate-operator-out.py for scoped enum classes.
Bug: 20845490
Change-Id: I6b67e288b1db563699161e58ec2e2330d42dd8f5
|
| 1f2d3ba6af52cf6f566deb38b7e07735c9a08fb6 |
|
18-May-2015 |
Mingyao Yang <mingyao@google.com> |
Fix nested deoptimization.
Handle nested deoptimization cases. Create a stacked shadow frame
records to keep track of deoptimization shadow frames. Shadow frames
under construction can be tracked in the same stack.
Bug: 20845490
Change-Id: I768285792c29e7c3cfcd21e7a2600802506024d8
|
| 3d21bdf8894e780d349c481e5c9e29fe1556051c |
|
22-Apr-2015 |
Mathieu Chartier <mathieuc@google.com> |
Move mirror::ArtMethod to native
Optimizing + quick tests are passing, devices boot.
TODO: Test and fix bugs in mips64.
Saves 16 bytes per most ArtMethod, 7.5MB reduction in system PSS.
Some of the savings are from removal of virtual methods and direct
methods object arrays.
Bug: 19264997
(cherry picked from commit e401d146407d61eeb99f8d6176b2ac13c4df1e33)
Change-Id: I622469a0cfa0e7082a2119f3d6a9491eb61e3f3d
Fix some ArtMethod related bugs
Added root visiting for runtime methods, not currently required
since the GcRoots in these methods are null.
Added missing GetInterfaceMethodIfProxy in GetMethodLine, fixes
--trace run-tests 005, 044.
Fixed optimizing compiler bug where we used a normal stack location
instead of double on ARM64, this fixes the debuggable tests.
TODO: Fix JDWP tests.
Bug: 19264997
Change-Id: I7c55f69c61d1b45351fd0dc7185ffe5efad82bd3
ART: Fix casts for 64-bit pointers on 32-bit compiler.
Bug: 19264997
Change-Id: Ief45cdd4bae5a43fc8bfdfa7cf744e2c57529457
Fix JDWP tests after ArtMethod change
Fixes Throwable::GetStackDepth for exception event detection after
internal stack trace representation change.
Adds missing ArtMethod::GetInterfaceMethodIfProxy call in case of
proxy method.
Bug: 19264997
Change-Id: I363e293796848c3ec491c963813f62d868da44d2
Fix accidental IMT and root marking regression
Was always using the conflict trampoline. Also included fix for
regression in GC time caused by extra roots. Most of the regression
was IMT.
Fixed bug in DumpGcPerformanceInfo where we would get SIGABRT due to
detached thread.
EvaluateAndApplyChanges:
From ~2500 -> ~1980
GC time: 8.2s -> 7.2s due to 1s less of MarkConcurrentRoots
Bug: 19264997
Change-Id: I4333e80a8268c2ed1284f87f25b9f113d4f2c7e0
Fix bogus image test assert
Previously we were comparing the size of the non moving space to
size of the image file.
Now we properly compare the size of the image space against the size
of the image file.
Bug: 19264997
Change-Id: I7359f1f73ae3df60c5147245935a24431c04808a
[MIPS64] Fix art_quick_invoke_stub argument offsets.
ArtMethod reference's size got bigger, so we need to move other args
and leave enough space for ArtMethod* and 'this' pointer.
This fixes mips64 boot.
Bug: 19264997
Change-Id: I47198d5f39a4caab30b3b77479d5eedaad5006ab
|
| e401d146407d61eeb99f8d6176b2ac13c4df1e33 |
|
22-Apr-2015 |
Mathieu Chartier <mathieuc@google.com> |
Move mirror::ArtMethod to native
Optimizing + quick tests are passing, devices boot.
TODO: Test and fix bugs in mips64.
Saves 16 bytes per most ArtMethod, 7.5MB reduction in system PSS.
Some of the savings are from removal of virtual methods and direct
methods object arrays.
Bug: 19264997
Change-Id: I622469a0cfa0e7082a2119f3d6a9491eb61e3f3d
|
| 8e5bd18fc665d7ec5461ea068e98740a65da754c |
|
06-May-2015 |
Nicolas Geoffray <ngeoffray@google.com> |
Add a flag to StackVisitor for inlining.
The flag tells whether the stack walk needs to include inlined
Java frames.
This does not do anything just yet, as we're not inlining anyways.
Change-Id: I716e25094fe56fa335ca1f9a398c1bcdba478e73
|
| 50030ef998be09789da4a9a56738362852068f12 |
|
08-May-2015 |
Mathieu Chartier <mathieuc@google.com> |
Check IsReferenceVReg during deopt
Required since the quick GC maps may not agree with the verifier
ones. Without this check we may copy stale object references into
the shadow frame.
(cherry picked from commit f00baf56ef647684888a407dbb6adadd704a2039)
Bug: 20736048
Change-Id: I7783c8a8ee45cf601b08b4c38f1dec7f7d11380c
|
| f00baf56ef647684888a407dbb6adadd704a2039 |
|
08-May-2015 |
Mathieu Chartier <mathieuc@google.com> |
Check IsReferenceVReg during deopt
Required since the quick GC maps may not agree with the verifier
ones. Without this check we may copy stale object references into
the shadow frame.
Bug: 20736048
Change-Id: I7783c8a8ee45cf601b08b4c38f1dec7f7d11380c
|
| 241a9588c6d7d0fcb4c92da40c7141863930083a |
|
28-Apr-2015 |
Christopher Ferris <cferris@google.com> |
Fix mismatched new[]/delete.
Another two cases where a new[] is used but only a delete occurs.
Bug: 18202869
Change-Id: If68264807150f3a9783e44ef8823cc366bff8df2
|
| 15b9d5274399736ac09705f0507df24fac4f00c1 |
|
12-Mar-2015 |
Nicolas Geoffray <ngeoffray@google.com> |
API change in StackVisitor::GetVReg*.
- Remove GetVReg() and SetVReg() that were expecting to always succeed.
- Change Quick-only methods to take a FromQuickCode suffix.
- Change deopt to use dead values when GetVReg does not succeed:
the optimizing compiler will not have a location for uninitialized
Dex registers and potentially dead registers.
Change-Id: Ida05773a97aff8aa69e0caf42ea961f80f854b77
|
| 14691c5e786e8c2c5734f687e4c96217340771be |
|
05-Mar-2015 |
Nicolas Geoffray <ngeoffray@google.com> |
Compute the right catch location for the debugger.
Also remove tls ThrowLocation, it is not needed anymore.
Change-Id: I78fddf09ce968ca475e39c17fa76d699c589c8d9
|
| 7642cfc90fc9c3ebfd8e3b5041915705c93b5cf0 |
|
26-Feb-2015 |
Nicolas Geoffray <ngeoffray@google.com> |
Change how we report exceptions to the debugger.
This is only a refactoring/cleanup. Bug fixes with respect
to catch location, and more cleanups will follow.
Change-Id: I30d3c6260b0c8f8115a811621397225b88f2063a
|
| 794ad76e8d5b5b9132819d5b08a0570e27615644 |
|
23-Feb-2015 |
Andreas Gampe <agampe@google.com> |
ART: Introduce NO_RETURN, Mark DoLongJump noreturn
Add NO_RETURN macro that adds C++11 noreturn attribute. Mark
DoLongJump methods as noreturn.
Change-Id: Ifde4318e370493237050d4c1349285a0382df23f
|
| 2e04bb2d514434f2bedb14268996842010948e6c |
|
11-Feb-2015 |
Andreas Gampe <agampe@google.com> |
ART: Allow class-loading during deopt
When deoptimizing, we might touch code that uses unloaded classes.
Bug: 19290147
(cherry picked from commit 44fb719e5f0f5ee7dcf4b1eae703593f1043a169)
Change-Id: I5776f08ba366e9742336caba0d6af85f00629afc
|
| 44fb719e5f0f5ee7dcf4b1eae703593f1043a169 |
|
11-Feb-2015 |
Andreas Gampe <agampe@google.com> |
ART: Allow class-loading during deopt
When deoptimizing, we might touch code that uses unloaded classes.
Bug: 19290147
Change-Id: I5776f08ba366e9742336caba0d6af85f00629afc
|
| 90a33595bc637f5768a7726a186bdfe25efcd0d6 |
|
16-Jan-2015 |
Sebastien Hertz <shertz@google.com> |
Fix exception handling during deoptimization
When interpreting a deoptimized shadow frame, we may start with a
pending exception thrown by a previous deoptimized shadow frame (from
a previous invoke). Therefore, we need to handle it before executing
any instruction, otherwise we execute incorrect code.
Because we need the DEX pc of the throwing instruction to find a
matching catch handler, we initialize deoptimized shadow frames with
the current DEX pc at the time the stack is deoptimized.
When we are about to interpret a deoptimized shadow frame, we need to
update the shadow frame with the DEX pc of the next instruction to
interpret. There are three cases:
- if there is no pending exception, this is the instruction following
the current one.
- if there is a pending exception and we found a matching catch
handler, this is the first instruction of this handler.
- if there is a pending exception but there is no matching catch
handler, we do not execute the deoptimized shadow frame and continue
to its caller.
The verifier now fails when a method starts with a move-exception
instruction. Indeed we cannot start executing a method with a pending
exception.
Bug: 19057915
Bug: 19041195
Bug: 18607595
(cherry picked from commit 270a0e16c3b8e5b95cbfdbd8996ac137c7c6322b)
Change-Id: Ib4fdd0ad704b4f2369d229737c9cc77f559cea55
|
| 270a0e16c3b8e5b95cbfdbd8996ac137c7c6322b |
|
16-Jan-2015 |
Sebastien Hertz <shertz@google.com> |
Fix exception handling during deoptimization
When interpreting a deoptimized shadow frame, we may start with a
pending exception thrown by a previous deoptimized shadow frame (from
a previous invoke). Therefore, we need to handle it before executing
any instruction, otherwise we execute incorrect code.
Because we need the DEX pc of the throwing instruction to find a
matching catch handler, we initialize deoptimized shadow frames with
the current DEX pc at the time the stack is deoptimized.
When we are about to interpret a deoptimized shadow frame, we need to
update the shadow frame with the DEX pc of the next instruction to
interpret. There are three cases:
- if there is no pending exception, this is the instruction following
the current one.
- if there is a pending exception and we found a matching catch
handler, this is the first instruction of this handler.
- if there is a pending exception but there is no matching catch
handler, we do not execute the deoptimized shadow frame and continue
to its caller.
The verifier now fails when a method starts with a move-exception
instruction. Indeed we cannot start executing a method with a pending
exception.
Bug: 19057915
Bug: 19041195
Bug: 18607595
Change-Id: I355ac81e6ac098edc7e3cc8c13dbfa24a2969ab2
|
| 4306ef8a7ec8e3887e51f64e80d940d974cc3ac3 |
|
20-Dec-2014 |
Mathieu Chartier <mathieuc@google.com> |
Don't allow suspension from FindLocksAtDexPc
Transitioning to suspended from runnable sometimes runs dump
checkpoints in ThreadStress which can cause deadlocks. This happens
since FindLocksAtDexPC runs the verifier which calls
AllowThreadSuspension. This results in a blocked thread which holds
the monitor lock, and if another thread tries to do a monitor enter,
it deadlocks while holding the mutator lock (assuming the GC is
suspending all).
The fix for avoiding this deadlock is not calling
AllowThreadSuspension from FindLocksAtDexPc.
Bug: 18576985
Change-Id: I7e5faaf3bbbd5b5f680de95d53c33b5106705b0c
|
| 6a3c1fcb4ba42ad4d5d142c17a3712a6ddd3866f |
|
31-Oct-2014 |
Ian Rogers <irogers@google.com> |
Remove -Wno-unused-parameter and -Wno-sign-promo from base cflags.
Fix associated errors about unused paramenters and implict sign conversions.
For sign conversion this was largely in the area of enums, so add ostream
operators for the effected enums and fix tools/generate-operator-out.py.
Tidy arena allocation code and arena allocated data types, rather than fixing
new and delete operators.
Remove dead code.
Change-Id: I5b433e722d2f75baacfacae4d32aef4a828bfe1b
|
| 20ffaaab1339bbda450f4ff4262910b5f34da085 |
|
29-Oct-2014 |
Sebastien Hertz <shertz@google.com> |
Fix deoptimization bugs
Only concerns long and double variables which use a register pair.
Bug: 18167243
(partial cherry-pick of commit 07140838a3ee44a6056cacdc78f2930e019107da)
Change-Id: Ibbbd9b1ed576a1502ca0025a3e27176014ed2548
|
| cf7f19135f0e273f7b0136315633c2abfc715343 |
|
23-Oct-2014 |
Ian Rogers <irogers@google.com> |
C++11 related clean-up of DISALLOW_..
Move DISALLOW_COPY_AND_ASSIGN to delete functions. By no having declarations
with no definitions this prompts better warning messages so deal with these
by correcting the code.
Add a DISALLOW_ALLOCATION and use for ValueObject and mirror::Object.
Make X86 assembly operand types ValueObjects to fix compilation errors.
Tidy the use of iostream and ostream.
Avoid making cutils a dependency via mutex-inl.h for tests that link against
libart. Push tracing dependencies into appropriate files and mutex.cc.
x86 32-bit host symbols size is increased for libarttest, avoid copying this
in run-test 115 by using symlinks and remove this test's higher than normal
ulimit.
Fix the RunningOnValgrind test in RosAllocSpace to not use GetHeap as it
returns NULL when the heap is under construction by Runtime.
Change-Id: Ia246f7ac0c11f73072b30d70566a196e9b78472b
|
| 677cd61ad05d993c4d3b22656675874f06d6aabc |
|
15-Oct-2014 |
Ian Rogers <irogers@google.com> |
Make ART compile with GCC -O0 again.
Tidy up InstructionSetFeatures so that it has a type hierarchy dependent on
architecture.
Add to instruction_set_test to warn when InstructionSetFeatures don't agree
with ones from system properties, AT_HWCAP and /proc/cpuinfo.
Clean-up class linker entry point logic to not return entry points but to
test whether the passed code is the particular entrypoint. This works around
image trampolines that replicate entrypoints.
Bug: 17993736
(cherry picked from commit 6f3dbbadf4ce66982eb3d400e0a74cb73eb034f3)
Change-Id: I3e7595f437db4828072589d475a5453b7f31003e
|
| 6f3dbbadf4ce66982eb3d400e0a74cb73eb034f3 |
|
15-Oct-2014 |
Ian Rogers <irogers@google.com> |
Make ART compile with GCC -O0 again.
Tidy up InstructionSetFeatures so that it has a type hierarchy dependent on
architecture.
Add to instruction_set_test to warn when InstructionSetFeatures don't agree
with ones from system properties, AT_HWCAP and /proc/cpuinfo.
Clean-up class linker entry point logic to not return entry points but to
test whether the passed code is the particular entrypoint. This works around
image trampolines that replicate entrypoints.
Bug: 17993736
Change-Id: I5f4b49e88c3b02a79f9bee04f83395146ed7be23
|
| 07140838a3ee44a6056cacdc78f2930e019107da |
|
01-Oct-2014 |
Ian Rogers <irogers@google.com> |
Enable -Wunreachable-code
Caught bugs in DeoptimizeStackVisitor and assemble_x86 SIB encoding.
Add UNREACHABLE macro to document code expected to be unreachable.
Bug: 17731047
Change-Id: I2e363fe5b38a1246354d98be18c902a6031c0b9e
|
| 7b078e8c04f3e1451dbdd18543c8b9692b5b067e |
|
10-Sep-2014 |
Ian Rogers <irogers@google.com> |
Compile time performance improvements focusing on interpret-only.
Reduce virtual method dispatch in the method verifier and make more code
inline-able.
Add a StringPiece with const char* equality operator to avoid redundant
StringPieces and strlens.
Remove back link from register line to verifier and pass as argument to reduce
size of RegisterLine.
Remove instruction length from instruction flags and compute from the
instruction, again to reduce size.
Add suspend checks to resolve and verify to allow for more easy monitor
inflation and reduce contention on Locks::thread_list_suspend_thread_lock_.
Change ThrowEarlierClassFailure to throw pre-allocated exception.
Avoid calls to Thread::Current() by passing self.
Template specialize IsValidClassName.
Make ANR reporting with SIGQUIT run using checkpoints rather than suspending
all threads. This makes the stack/lock analysis less lock error prone.
Extra Barrier assertions and condition variable time out is now returned as a
boolean both from Barrier and ConditionVariable::Wait.
2 threaded host x86-64 interpret-only numbers from 341 samples:
Before change: Avg 176.137ms 99% CI 3.468ms to 1060.770ms
After change: Avg 139.163% 99% CI 3.027ms to 838.257ms
Reduction in average compile time after change is 20.9%.
Slow-down without change is 26.5%.
Bug: 17471626 - Fix bug where RegTypeCache::JavaLangObject/String/Class/Throwable
could return unresolved type when class loading is disabled.
Bug: 17398101
Change-Id: Id59ce3cc520701c6ecf612f7152498107bc40684
|
| bf99f77dda749e2b653e8c45259b1fb56e7bb012 |
|
24-Aug-2014 |
Mathieu Chartier <mathieuc@google.com> |
Clean up Handle usage.
Prefer using ConstHandle instead of Handle as function arguments
since you can't assign new references to ConstHandle which helps
prevent bugs.
Changed NullHandle to be a ConstHandle so that you can never modify
it to be a non null reference.
Change-Id: I81cb979f6f8d5b49e5614966a2caf28c3701dd4f
|
| dc3761719fb5e2d1ced5708e3c73b965f9ef0c73 |
|
22-Aug-2014 |
Hiroshi Yamauchi <yamauchi@google.com> |
Handlerize MethodVerifier::mirror_method_.
The method (mirror_method_) wasn't handlerized across some
allocation/GC points such as the ResolveType() call in
ScanTryCatchBlocks() and the GetReturnType() calls in
CodeFlowVerifyInstruction().
Bug: 12687968
Change-Id: I95323de14459eb5a7c4abfcf44f882f86d59be64
|
| 2e981cbaa2e869ec3910ebdc5cce9d4ab9edb758 |
|
13-Aug-2014 |
Hiroshi Yamauchi <yamauchi@google.com> |
More efficient stack walk in exception throwing.
In the exception handling code, we currently walk down the stack
twice, once to get the stack height which we use to compute frame IDs
(the bottom frame is zero), and once more to find the catch block to
jump to.
For a deep stack, this could result in very slow exception
handling. That is, if have a lot of finally or catch blocks that we
end up jumping to in a deep stack, we need to do a lot of
catch/rethrow chains. Since we'd need to walk down to the bottom each
time to compute frames IDs in each catch/rethrow, we'd need to walk
down O(N^2) frames at the worst case.
Instead of frames IDs ((the bottom frame is zero), we will use the
frame depth (the top frame is zero) and no longer need to walk down
the stack just to get the stack height. We walk down O(N) frames.
This was what was happening with
code.google.gson.functional.CircularReferenceTest. With this change,
the test run time went from ~120s down to ~3s on N5 and it no longer
crashes due to the thread suspension timeout.
(cherry pick commit 649278cec7119cdd1bea3d0b710dbb2aa7c650b6)
Bug: 16800209
Change-Id: Ie815df1e3e8fb9d82e40685d4cc2b8838fd8aa07
|
| 649278cec7119cdd1bea3d0b710dbb2aa7c650b6 |
|
13-Aug-2014 |
Hiroshi Yamauchi <yamauchi@google.com> |
More efficient stack walk in exception throwing.
In the exception handling code, we currently walk down the stack
twice, once to get the stack height which we use to compute frame IDs
(the bottom frame is zero), and once more to find the catch block to
jump to.
For a deep stack, this could result in very slow exception
handling. That is, if have a lot of finally or catch blocks that we
end up jumping to in a deep stack, we need to do a lot of
catch/rethrow chains. Since we'd need to walk down to the bottom each
time to compute frames IDs in each catch/rethrow, we'd need to walk
down O(N^2) frames at the worst case.
Instead of frames IDs ((the bottom frame is zero), we will use the
frame depth (the top frame is zero) and no longer need to walk down
the stack just to get the stack height. We walk down O(N) frames.
This was what was happening with
code.google.gson.functional.CircularReferenceTest. With this change,
the test run time went from ~120s down to ~3s on N5 and it no longer
crashes due to the thread suspension timeout.
Bug: 16800209
Change-Id: Ie815df1e3e8fb9d82e40685d4cc2b8838fd8aa07
|
| 5dc158e9b6635b6c5d3916ae4094e8886feb4580 |
|
16-Jul-2014 |
Sebastien Hertz <shertz@google.com> |
Fix access to long/double stack values from debugger
Long and double values live in a pair of DEX registers. When we compile DEX
code with the Quick compiler, a DEX register either lives in the stack or is
promoted to a physical register. In the case of a pair of DEX registers, the
Quick compiler assumes both registers live in the same "area": both live in
the stack or both are promoted to physical registers.
From the debugger, we used to access these values by reading/writing each DEX
register separately. However, this does not work when only one DEX register of
a pair is promoted and the other lives in the stack. In this case, the compiled
code reads from/writes to the stack only.
To fix that, the debugger must follow the same rule than the Quick compiler: if
both DEX registers are promoted, read/write them from/to physical registers,
otherwise read/write them from/to the stack. We add StackVisitor:GetVRegPair and
StackVisitor:SetVRegPair for this purpose.
We also follow the same rule when deoptimizing. However we need to do that only
when we know two consecutive DEX registers are part of a pair (long or double).
We know that thanks to the verifier.
Bug: 15527793
(cherry picked from commit c901dd7bdc80b953d04100ef2f54b8d1ca5f466b)
Change-Id: I981e088295254d75352f83a3e6ba0b292cfe3ed3
|
| c901dd7bdc80b953d04100ef2f54b8d1ca5f466b |
|
16-Jul-2014 |
Sebastien Hertz <shertz@google.com> |
Fix access to long/double stack values from debugger
Long and double values live in a pair of DEX registers. When we compile DEX
code with the Quick compiler, a DEX register either lives in the stack or is
promoted to a physical register. In the case of a pair of DEX registers, the
Quick compiler assumes both registers live in the same "area": both live in
the stack or both are promoted to physical registers.
From the debugger, we used to access these values by reading/writing each DEX
register separately. However, this does not work when only one DEX register of
a pair is promoted and the other lives in the stack. In this case, the compiled
code reads from/writes to the stack only.
To fix that, the debugger must follow the same rule than the Quick compiler: if
both DEX registers are promoted, read/write them from/to physical registers,
otherwise read/write them from/to the stack. We add StackVisitor:GetVRegPair and
StackVisitor:SetVRegPair for this purpose.
We also follow the same rule when deoptimizing. However we need to do that only
when we know two consecutive DEX registers are part of a pair (long or double).
We know that thanks to the verifier.
Bug: 15527793
Change-Id: I04812285ff26ef0129f39792a1cf776f3591ca2d
|
| e63db27db913f1a88e2095a1ee8239b2bb9124e8 |
|
16-Jul-2014 |
Ian Rogers <irogers@google.com> |
Break apart header files.
Create libart-gtest for common runtime and compiler gtest routines.
Rename CompilerCallbacksImpl that is quick compiler specific.
Rename trace clock source constants to not use the overloaded profiler term.
Change-Id: I4aac4bdc7e7850c68335f81e59a390133b54e933
|
| 98d1cc8033251c93786e2fa8c59a2e555a9493be |
|
16-May-2014 |
Mingyao Yang <mingyao@google.com> |
Improve performance of invokevirtual/invokeinterface with embedded imt/vtable
Add an embedded version of imt/vtable into class object. Both tables start at
fixed offset within class object so method/entry point can be loaded directly
from class object for invokeinterface/invokevirtual.
Bug: 8142917
Change-Id: I4240d58cfbe9250107c95c0708c036854c455968
|
| 9f1020305292a21fd14a402b189c765a125226ab |
|
23-May-2014 |
Sebastien Hertz <shertz@google.com> |
Fix exception reporting from interpreter
To comply with JDWP exception report rules, we must report an exception at the
location of the throw (or the first instruction encountered after a native
call). To do this, we use the CatchLocationFinder visitor to look for a catch
handler until we reach a native frame or the top frame.
Because interpreter handles pending exception on a method-by-method basis, we
need a flag to remember we already reported the exception and avoid reporting
it multiple times when unwinding methods. The drawback is we need to maintain
the state of this flag. We clear it when the exception is cleared. In the case
we temporarily clear the exception (when finding a catch handler for instance),
we restore the flag to its previous value at the same time we restore the
pending exception.
Bump oat version to force recompilation because we modify Thread offsets.
Bug: 14402770
Change-Id: Ic059c58f80b2023b118038301f8f0a24f1e18241
|
| bfd9a4378eacaf2dc2bbe05ad48c5164fc93c9fe |
|
22-May-2014 |
Mathieu Chartier <mathieuc@google.com> |
Change MethodHelper to use a Handle.
Added ConstHandle to help prevent errors where you modify the value
stored in the handle of the caller. Also fixed compaction bugs
related to not knowing MethodHelper::GetReturnType can resolve types.
This bug was present in interpreter RETURN_OBJECT.
Bug: 13077697
Change-Id: I71f964d4d810ab4debda1a09bc968af8f3c874a3
|
| ffddfdf6fec0b9d98a692e27242eecb15af5ead2 |
|
03-Jun-2014 |
Tim Murray <timmurray@google.com> |
DO NOT MERGE
Merge ART from AOSP to lmp-preview-dev.
Change-Id: I0f578733a4b8756fd780d4a052ad69b746f687a9
|
| 5cf98196d488437acd1e989c08a554ef697fded1 |
|
30-May-2014 |
Ian Rogers <irogers@google.com> |
Don't report down-calls as unhandled exceptions.
Bug: 15310540
Also, narrow scope of catch/deoptimize stack visitors that are specific to
quick exception delivery.
Change-Id: Ib13a006ce1347acb93a36b0186550d4c3ec2034b
|
| cf4035a4c41ccfcc3e89a0cee25f5218a11b0705 |
|
29-May-2014 |
Andreas Gampe <agampe@google.com> |
ART: Use StackReference in Quick Stack Frame
The method reference at the bottom of a quick frame is a stack
reference and not a native pointer. This is important for 64b
architectures, where the notions do not coincide.
Change key methods to have StackReference<mirror::ArtMethod>*
parameter instead of mirror::ArtMethod**. Make changes to
invoke stubs for 64b archs, change the frame setup for JNI code
(both generic JNI and compilers), tie up loose ends.
Tested on x86 and x86-64 with host tests. On x86-64, tests succeed
with jni compiler activated. x86-64 QCG was not tested.
Tested on ARM32 with device tests.
Fix ARM64 not saving x19 (used for wSUSPEND) on upcalls.
Tested on ARM64 in interpreter-only + generic-jni mode.
Fix ARM64 JNI Compiler to work with the CL.
Tested on ARM64 in interpreter-only + jni compiler.
Change-Id: I77931a0cbadd04d163b3eb8d6f6a6f8740578f13
|
| eb8167a4f4d27fce0530f6724ab8032610cd146b |
|
08-May-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add Handle/HandleScope and delete SirtRef.
Delete SirtRef and replaced it with Handle. Handles are value types
which wrap around StackReference*.
Renamed StackIndirectReferenceTable to HandleScope.
Added a scoped handle wrapper which wraps around an Object** and
restores it in its destructor.
Renamed Handle::get -> Get.
Bug: 8473721
Change-Id: Idbfebd4f35af629f0f43931b7c5184b334822c7a
|
| ff093b31d75658c3404f9b51ee45760f346f06d9 |
|
01-May-2014 |
Ian Rogers <irogers@google.com> |
Fix a few 64-bit compilation of 32-bit code issues.
Bug: 13423943
Change-Id: I939389413af0a68c0d95b23cd598b7c42afa4383
|
| 714f175bd66d03225927a84f3d5dbc923c5a3e7e |
|
28-Apr-2014 |
Sebastien Hertz <shertz@google.com> |
Visit deoptimized shadow frames as roots
During deoptimization, we create shadow frames but do not attach them to the
stack until we transition to interpreter mode. If a GC happens before that,
these shadow frames are not visited by GC so they may contain stale references.
This CL addresses this issue by visiting the shadow frames "under contruction"
in Thread::VisitRoots so we correctly update all references they hold.
To make them visible, we now save the top shadow frame (the first one created)
in the field Thread::tls_ptr_sized_values::deoptimization_shadow_frame. This
field will then be cleared when transitioning to interpreter mode.
Bug: 14324885
Change-Id: Ib213ddc80f19cfcdfcec6f20acaa7f1a0e9ce2c1
|
| fd3077e4b9ebadd281777310d26e64443858f653 |
|
23-Apr-2014 |
Sebastien Hertz <shertz@google.com> |
Refactor exception handling for deoptimization
This CL refactors the exception handling (on the quick side) by isolating the
search of catch handler and the preparation of deoptimization.
We rename the CatchFinder class to QuickExceptionHandler so it's less specific
to catch handler search.
Finding catch handler happens in QuickExceptionHandler::FindCatch. Since the
CatchBlockStackVisitor resolves exception types, it may cause thread suspension
and breaks the assertion current thread can't be suspended. Therefore, we place
the exception in a SirtRef (while it is detached from the current thread) and
remove the thread suspension assertion.
Deoptimization now happens in QuickExceptionHandler::DeoptimizeStack. It uses
the new DeoptimizeStackVisitor class to create shadow frames.
We also add the Thread::GetDeoptimizationException method to get the definition
of the fake exception in only one place.
Change-Id: I01b19fa72af64329b5c3b6c7f0c3339d2d724978
|