New folder structure (nested under vm/):
- compiler/
- jit/ - JIT specific code
- aot/ - AOT specific code
- backend/ - all middle-end and back-end code (IL, flow graph)
- assembler/ - assemblers and disassemblers
- frontend/ - front ends (AST -> IL, Kernel -> IL)
compiler/README.md would be the documentation root for the compiler
pipeline
Bug: https://github.com/dart-lang/sdk/issues/30575
Change-Id: I2dfd9688793bff737f7632ddc77fca766875ce36
Reviewed-on: https://dart-review.googlesource.com/2940
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Commit-Queue: Vyacheslav Egorov <vegorov@google.com>
Use that to sync variables to predefined stack
slots inside of exception handler, not before
each MayThrow() instruction.
In JIT mode we can avoid storing extra metadata
by using deopt info instead.
Introduce caching of metadata and
exception handler based on PC.
R=fschneider@google.com, vegorov@google.com
Review-Url: https://codereview.chromium.org/2734323003 .
Use that to sync variables to predefined stack
slots inside of exception handler, not before
each MayThrow() instruction.
In JIT mode we can avoid storing extra metadata
by using deopt info instead.
Introduce caching of metadata and
exception handler based on PC.
R=fschneider@google.com
Review-Url: https://codereview.chromium.org/2640853002 .
i.e. #ifndef VM_WHATEVER -> #ifndef RUNTIME_VM_WHATEVER
This lets us remove a hack from the PRESUBMIT.py script that existed
for reasons that are no longer valid, and sets us up to add some
presubmit checks for the GN build.
R=asiva@google.com, rmacnak@google.com
Review URL: https://codereview.chromium.org/2450713004 .
This CL enables unboxed doubles for simdbc64. Unboxed
doubles are stored in regular dbc registers, which are
really 64-bit stack slots. It also implements binary
and unary double operations and comparisons.
R=vegorov@google.com
Review URL: https://codereview.chromium.org/2120703002 .
Most of the infrastructure is fixed to work with DBC stack layout:
- register allocator allocates DBC registers with the limitation that we allocate only 20 registers and bail out if anything needs spilling (there is no use implementing spilling on DBC because registers are memory locations themselves). We should be able to bump number of CPU registers on DBC up to 256 but this requires major surgery in some parts - so I postponed this;
- lazy deoptimization is implemented, eager deoptimization is not - because we don't emit any code that actually requires it. it's a minor change to support it once we have a target;
- stack scanning respects stack maps built by registers allocator;
We bailout from all unsupported instructions.
R=zra@google.com
Review URL: https://codereview.chromium.org/1992963002 .
Instead of calling code object directly, call indirectly and
pass the code object in a register. The object pool is then loaded from
the code object. This is another preparation step for making generated code
relocatable.
All non-ia32 platforms:
No entry patching.
ARM:
PC marker (now code object) moves to the same place as on x64 (below saved PP, above saved FP).
R9 is now used as PP, R10 as CODE_REG.
BUG=
R=rmacnak@google.com
Review URL: https://codereview.chromium.org//1192103004 .
- Each isolate has its own TimelineEventBuffer
- Each isolate has TimelineEventStreams for API, Compiler, Embedder, Isolate, and GC
- Compiler, Isolate, Embedder, and GC are being inserted into the stream
- Basic unit tests
- Dart API for injecting timeline events
- Arbitrary number of native and Dart arguments can be attached to each event
- JSON printing
- Output can be loaded by about://tracing
- Add --timeline-trace-dir flag
Short term follow up CLS:
- Service protocol requests to enable / disable tracing and retrieve a trace
- UI for Observatory
- Dart code EventStreams and Events
R=asiva@google.com
Review URL: https://codereview.chromium.org//1170503004.
This adds meta-information to object pool entries to allow storing
untagged immediates or code addresses (ExternalLabel) directly.
This eliminates the need to generate extra code to preserve the LSB
when storing immediates as smis (x64, arm64).
BUG=
Review URL: https://codereview.chromium.org//1175523002.
Add deferred deoptimization instructions for return address, pc marker and
object pool pointer. This enables generating unoptimized code lazily
on deoptimization.
This CL does not change the way unoptimized code is generated or collected
yet, but just provides the mechanisms needed. This can be used to allow
code GC for the unoptimized code of optimized functions.
R=iposva@google.com
Review URL: https://codereview.chromium.org//1052563003
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@44934 260f80e4-7a28-3924-810f-c04153c831b5
Refactor all remaning cases where the current zone is used through new(Isolate*) and remove this interface.
Removing this interface is needed to move towards multiple threads per isolate, and also makes the caller more aware of the scope of the zone used, reducing the risk of use-after-free.
Make the current thread and the stack zone created around native/runtime entries directly available in their body, saving an indirection (and optimized away if unused).
R=iposva@google.com
Review URL: https://codereview.chromium.org//982873004
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@44541 260f80e4-7a28-3924-810f-c04153c831b5
replace it with a sequence of checks that guarantee
0 <= LowerBound(index) < UpperBound(index) < length
and hoist all of those checks out of the enclosing loop.
Upper/Lower bounds are symbolic arithmetic expressions with +, -, *
operations and are computed based on discovered simple induction variables.
Simple induction variable is a variable that follows the pattern v1 <- phi(v0, v1 + 1)
BUG=
R=fschneider@google.com
Review URL: https://codereview.chromium.org//619903002
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@40969 260f80e4-7a28-3924-810f-c04153c831b5
- Improve escape/alias analysis:
-- Storing an object into a field of another object does not mean that this object escapes (or has aliases) as long as that object does not have any loads from the same place;
-- Places like X.f and Y.f don't alias if X and Y are two different allocation instructions even if X and Y themselves potentially have aliases;
-- Improve precision of alias analysis for indexed properties;
- Support dematerialization and rematerialization of objects that are referenced by other dematerialized objects.
-- Use fix-point algorithm to collect candidates for allocation sinking;
-- Support aborting unsuccessful allocation sinking.
R=fschneider@google.com, johnmccutchan@google.com
Review URL: https://codereview.chromium.org//395943003
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@38404 260f80e4-7a28-3924-810f-c04153c831b5
Changes:
* Register allocator now allocates GPRs for kUnboxedMint.
* Register allocator supports for SameAsFirstInput for register pairs.
* Register allocator properly handles register pairs in environment uses and materialization uses.
* BoxInteger updated on IA32/ARM.
* UnboxInteger updated on IA32/ARM.
* BinaryMintOp updated on IA32/ARM.
* ShiftMintOp updated on IA32/ARM.
* UnaryMintOp updated on IA32/ARM.
* RelationalOp updated on IA32/ARM.
* EqualityCompare updated on IA32/ARM.
* LoadIndexed and StoreIndexed updated on IA32/ARM.
* New Deopt instructions added.
* Update live_registers when an instruction has a fixed register input and a call on the slow path.
* Improve printing of register pairs in flow graph.
* Do not assume live registers in slow paths contain tagged values.
* LiveRange pairs for kUnboxedMint definitions marked as kUntagged representation (reduces stack usage).
* Live register spilling on ARM uses same register order as stack map encoding.
* Spill slots containing tagged and untagged are segregated.
* Print stack maps when printing live ranges with safe points.
* Print allocated spill slot when printing live ranges.
Status:
* IA32 completed. All tests are passing.
* ARM completed. All tests passing.
R=fschneider@google.com, srdjan@google.com, zra@google.com
Review URL: https://codereview.chromium.org//252333002
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@36468 260f80e4-7a28-3924-810f-c04153c831b5
inlined frames.
------------
Most remaining deoptimization code from the code generator moves to
DeoptContext. This allows the code to be reused by the debugger.
There is some rework of the code along the way. The remaining code in
the code generator is simpler.
Implement the ability to deopt a frame to an Array. Each inlined
frame accesses its locals from this array at some fixed offset.
Refactor the Debugger::CollectStackTrace code. New code is int
Debugger::CollectStackTraceNew. There is a flag --use_new_stacktrace
which can be used to revert back to the old version. I intend to
remove this flag shortly, after any dust clears.
Added a unit test which makes sure that we can inspect locals from
optimized and inlined frames. Tested this code in the dart editor
debugger as well.
R=iposva@google.com, srdjan@google.com
Review URL: https://codereview.chromium.org//26255004
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@28468 260f80e4-7a28-3924-810f-c04153c831b5
Primarily this change moves deoptimization state/code out of Isolate
and into DeoptContext (formerly DeoptimizationContext). The lifetime
of DeoptizationContext changes to survive through the entire
deoptimization process.
Some minor renaming. DeoptizationContext -> DeoptContext to make it
consistent with DeoptInstr and to save my wrists.
New files deferred_object.{cc,h} contain a bunch of the stuff lifted
from isolate.{cc,h}.
R=fschneider@google.com, srdjan@google.com
Review URL: https://codereview.chromium.org//24834002
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@28112 260f80e4-7a28-3924-810f-c04153c831b5
When compressing the deoptimization info, suffixes of length one are not
shared (the compression is the same length as the original, but more
indirect). However, they were grafted onto the trie as if they were shared,
i.e., one level lower than they should be.
Add debug code to verify the compression/decompression round trip.
BUG=dartbug.com/13273
R=fschneider@google.com
Review URL: https://codereview.chromium.org//23788012
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@27602 260f80e4-7a28-3924-810f-c04153c831b5
AllocationSinking pass discovers non-escaping allocations that have no input uses other than uses in the stores into its own fields.
Every environment use of such allocation is replaced by a state snapshot (MaterializeObject instruction) that describes the state of each initialized field in the object. State snapshots are computed through an additional round of load-forwarding.
Once snapshots are computed allocations are removed from the graph.
MaterializeObject instructions are not compiled into native code but produce deoptimization instructions instead that describe how object should be materialized at deoptimization.
Deoptimization instructions now follow the following format:
[mat obj #1]...[mat obj #N][ret addr][... mat arguments ...][... real frames ...]
- the prefix describes each object to materialize on deopt via kMaterializeObject instruction;
- actual values that are needed for materialization are emited as a part of bottom-most stack frame. This is done to simplify implementation: they need to be discoverable by a GC during materialization phase. At the end of deoptimization they will be removed from the stack;
- normal stack slots can refer to materialized objects via kMaterializedObjectRef instruction.
Additionally this change contains fixes in load-forwarding that are needed to guarantee that all artificial LoadField instructions inserted during AllocationSinking are correctly replaced with actual values.
Limitations of the current implementation:
- can't eliminate allocations that flow into phis but otherwise don't actually escape;
- can't sink allocations out of loops;
- allocation with type arguments are not handled.
R=regis@google.com, srdjan@google.com, zra@google.com
Review URL: https://codereview.chromium.org//14935005
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@22485 260f80e4-7a28-3924-810f-c04153c831b5
Remove representation from location. Presence of representation in location encoding was violating the invariant that unequal locations must be disjoint (where equality for locations is defined in terms of bitwise equality of their encoding). This could lead ParallelMoveResolver to treat XMM1 containing unboxed double as unequal location to XMM1 containing unboxed mint, which is obviously incorrect.
For similar reason eliminate kFloat32x4StackSlot and kUint32x4StackSlot distinction is eliminated and both are replaced with kQuadStackSlot. Register allocator now guarantees that no kQuadStackSlot occupies the same space as any other kDoubleStackSlot. This also shrinks optimized stack when only doubles are used (but might lead to a higher stack utilization when a mixture of doubles and quads is used).
Implement allocation of scratch Cpu and Xmm registers for ParallelMoveResolver. This also allows to remove push(eax)/pop(eax) pairs when resolving memory-memory cycles on ia32.
BUG=dart:9710
Review URL: https://codereview.chromium.org//13801014
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@21148 260f80e4-7a28-3924-810f-c04153c831b5
Vyacheslav Egorov