Commit Graph

350 Commits

Author SHA1 Message Date
sgjesse@chromium.org
1e8079fcc9 Increase coverage of global loads in optimized code
In the cases where a global property cell cannot be used in the optimized code
use standard load ic to get the property instead of bailing out.

This is re-committing r7212 and r7215 which where reverted in r7239 with the addition of recoring the source position in the hydrogen code for the LoadGlobalCell instruction. To record that position an optional position field has been added to the variable proxy AST node.
Review URL: http://codereview.chromium.org/6758007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7474 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-04-01 11:54:04 +00:00
ricow@chromium.org
b4bae54ef2 Decrease number of runs in mjsunit/compiler/pic.js (Test still gets optimized, even without --stress-opt)
Review URL: http://codereview.chromium.org/6731049

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7414 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-03-29 13:20:05 +00:00
ricow@chromium.org
e0c7d3cf08 Run array-length less times to make the builder green.
We still get this method optimized even without --always-opt flag.
Review URL: http://codereview.chromium.org/6719029

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7381 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-03-28 08:51:08 +00:00
fschneider@chromium.org
e6cbf659d1 Fix bug that caused invalid code motion for certain loads instructions.
The dependency flags of instructions depending on a previous check have to
be a super-set of the flags of the check instructions.
Review URL: http://codereview.chromium.org/6730025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7343 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-03-24 11:37:24 +00:00
vitalyr@chromium.org
e26ae48786 Remove empty test/mjsunit/compiler/global-accessors.js to make lint happy.
Review URL: http://codereview.chromium.org/6712001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7244 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-03-17 16:12:13 +00:00
ricow@chromium.org
d6caa8872a Revert revisions 7215 and 7212.
This caueses line positions to be off by one in certain cases, causing webkit http/tests/inspector/console-xhr-logging to fail.
Review URL: http://codereview.chromium.org/6667077

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7239 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-03-17 14:30:48 +00:00
sgjesse@chromium.org
1a6c821b05 Increase coverage of global loads in optimized code
In the cases where a global property cell cannot be used in the optimized code use standard load ic to get the property instead of bailing out.
Review URL: http://codereview.chromium.org/6665026

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7212 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-03-17 08:16:12 +00:00
fschneider@chromium.org
031062d246 Fix bug with input representation of HValueOf.
The class did not correctly implement the RequiredInputRepresentation.
I changed this functions to be abstract so that all hydrogen classes
must implement it.

As a convention instructions with zero input operands return None as input
representation.

Instructions that can handle all input representations without converting before
also have None as required input representation (e.g. HTest)

All other instructions need a proper required input representation.

Review URL: http://codereview.chromium.org/6538088

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6891 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-02-22 12:14:53 +00:00
kmillikin@chromium.org
dc91c4218b Make optimized Function.prototype.apply safe for non-JSObject first arguments.
If we have a property access of the form this.x, where the access site sees
the global object, we can specialize the IC stub so that it performs a map
check without first performing a heap object check.

Ensure that we do not get in JS code with a non-JSObject this value by
deoptimizing at Function.prototype.apply if the first argument is not a
JSObject.

BUG=v8:1128

Review URL: http://codereview.chromium.org/6463025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6707 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-02-09 16:43:23 +00:00
lrn@chromium.org
2d15eb9a90 Fix bug in object literals with large array indexes as strings.
Review URL: http://codereview.chromium.org/6410028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6577 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-02-02 14:02:58 +00:00
fschneider@chromium.org
4e7ddab6dc Fix a bug in the placement of minus-zero checks and in GVN.
1. The placement of checks for negative zero has to be computed after
all conversion instructions have been inserted. I separated the code
into its own phase.

2. GVN need to take instruction flags into account when comparing
instructions for redundancy.

Review URL: http://codereview.chromium.org/6260035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6534 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-01-31 12:36:54 +00:00
antonm@chromium.org
49144ee6b6 Properly create variables to access outer arguments and function names.
Review URL: http://codereview.chromium.org/6266007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6380 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-01-19 08:16:17 +00:00
antonm@chromium.org
fae90d4f32 Make closures optimizable by Crankshaft compiler.
Currently only closures which only read from the context are supported.

Review URL: http://codereview.chromium.org/5753005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6340 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-01-17 08:11:03 +00:00
whesse@chromium.org
50eaaf4be4 Speed up a regression test regress-3249650.js so it does not timeout.
Review URL: http://codereview.chromium.org/6175001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6228 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2011-01-07 13:52:36 +00:00
ricow@chromium.org
717153274f Reduce number of iterations in new regression test, this times out on arm and the regression is hit with an order of magnitude less iterations.
Review URL: http://codereview.chromium.org/5963001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6058 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-12-17 07:23:06 +00:00
fschneider@chromium.org
73737fcdb6 Fix bugs in the range analysis for integers.
The overflow conditions were not correctly detected for 
certain add, sub and mul instructions.

I replaced the previous code by using 64-bit arithmetic
to correctly identify overflows for *, + and -.

Review URL: http://codereview.chromium.org/5860009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6055 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-12-16 18:01:36 +00:00
kmillikin@chromium.org
e0d3f6afdf Be more careful about exiting inlined functions in a test context.
When falling off the end of a function inlined in a test context, we cannot
constant fold the test of undefined away. The graph builder assumes that
control flow always reaches both branches of a test.

Instead, explicitly test and branch on "undefined". Introduce a pair of
empty blocks to hold the necessary LeaveInlined instructions.

Review URL: http://codereview.chromium.org/5566005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5927 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-12-07 12:07:40 +00:00
kasperl@chromium.org
90b3370374 Update V8 to version 3.0 (re-land r5920).
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5922 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-12-07 11:31:57 +00:00
kasperl@chromium.org
51b494d096 Revert r5920. Will re-land shortly.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5921 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-12-07 11:01:02 +00:00
kasperl@chromium.org
e5860bd6a8 Update V8 to version 3.0.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5920 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-12-07 09:11:56 +00:00
lrn@chromium.org
ecf0d85a8e Fix off-by-one in hex-parsing.
Review URL: http://codereview.chromium.org/5129002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5855 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-11-18 13:50:21 +00:00
whesse@chromium.org
ac60f498d2 Change keyed store IC interface on x64 to take value, key, and receiver in registers rather than on the stack.
Review URL: http://codereview.chromium.org/2111011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4692 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-05-20 15:09:21 +00:00
fschneider@chromium.org
03089b6644 Fix code generation for fast smi loops to support parameters as well.
This change fixes an assert we hit when we recognized a fast smi loop
with a parameter as the loop variable.

BUG=650


Review URL: http://codereview.chromium.org/1138003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4202 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-03-22 11:55:12 +00:00
fschneider@chromium.org
ec974ddb50 Fix bug when generating a fast smi loop.
We may encounter an invalid frame after generating code
for the loop body in case the loop body ends in an unconditional
return. Before setting the type information for the loop variable
we need to check for a valid frame.

Review URL: http://codereview.chromium.org/1106002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4182 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-03-18 14:32:02 +00:00
fschneider@chromium.org
34faea85db Add test cases for fast smi loops.
Review URL: http://codereview.chromium.org/1014007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4174 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-03-18 09:08:24 +00:00
fschneider@chromium.org
e090f46a55 Fix type propagation rules for count operation.
Also treat const-variables as not side-effect free.

Review URL: http://codereview.chromium.org/982009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4172 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-03-17 16:39:29 +00:00
fschneider@chromium.org
9202e05016 Fix bug in the count operation where we statically know the input is a smi.
Even if we know that the input to a count operation is a smi we still need to check if the result overflowed (and becomes a heap number).

Also fix the smi loop analysis to take two border cases correctly into account.

Review URL: http://codereview.chromium.org/1040002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4147 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-03-16 16:03:40 +00:00
whesse@chromium.org
b9a250d2e6 Fix error in compound assignment to keyed load by making platform-independent full compiler code platform dependent, add test of compound assignments.
Review URL: http://codereview.chromium.org/646009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3897 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-02-18 12:59:41 +00:00
fschneider@chromium.org
4a2c81d3b0 Add last use data flow information to the fast code generator.
This change add simple local live variable information to 
the fast code generator.  It supports only AST nodes that 
are accepted by the syntax checker.

Each variable use points to a variable definition structure
which contains the last use of the definition.

To determine whether a variable is live after a certain point
we can check whether its last use occurs later in the evaluation
order defined by the AST labeling number.

The new information is currently only printed out together with
the IR and not yet used for code generation.

Review URL: http://codereview.chromium.org/603004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3839 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-02-12 10:16:30 +00:00
kmillikin@chromium.org
b04a26dc31 Harden global variable accesses in the fast code generator.
Explicitly check that global variables do not have accessors or
interceptors in the fast code generator syntax checker.

Review URL: http://codereview.chromium.org/595022

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3828 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-02-10 13:46:15 +00:00
kmillikin@chromium.org
30cf88af6b Restrict the syntax that we aggressively optimize.
Do not use the speculative compiler for functions with other than one
statement in the body, and do not use it if subexpressions can have
side effects.  Bailing out to the beginning of the full code is not
sound if side effects have already occurred.

Add tests that would fail without the restrictions.

Review URL: http://codereview.chromium.org/598016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3826 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-02-10 09:18:55 +00:00
kmillikin@chromium.org
9b5c312bfc Initial implementation of fast path operation for bitwise OR.
Support a binary operation (bitwise OR) so long as it's not nested in
the left subexpression.  This ensures that the expression stack never
has height greater than two and so can be kept fully in registers.

The bounded expression stack height and the absence of any side
effects on the fast path allows us to still bailout out to the very
beginning of the function if any of our fast-path checks fail.

Review URL: http://codereview.chromium.org/594008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3822 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-02-09 13:44:43 +00:00
fschneider@chromium.org
2f0f6face1 Fast compiler: Load globals variables directly from property cells.
This is a first step towards loading globals directly from property cells instead
of going through a load IC.

This change supports only properties with the DontDelete attribute since
we are only able to bailout into the generic code generated by the secondary 
code generator the beginning of a function. The resulting fast-case code is 
specialized for a specific context. When invoked with a different global object, 
it will always bailout to the secondary code.

When loading a property that does not exist at compile-time or a property
that is deleteable we still generate the generic load IC.

Review URL: http://codereview.chromium.org/565034

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3808 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-02-05 13:57:18 +00:00
fschneider@chromium.org
58a8a1f341 Rename flags in unit tests to match the new flag definitions.
TBR=kmillikin@chromium.org, 
Review URL: http://codereview.chromium.org/555132

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3724 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-01-27 17:21:48 +00:00
ager@chromium.org
400b62a1f7 Implement unary addition in the full (non-optimizing) code generator.
Review URL: http://codereview.chromium.org/555099

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3711 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-01-26 13:57:11 +00:00
kmillikin@chromium.org
9095abb776 Introduce 'top-of-stack caching' to the toplevel code generator by
allowing the value of the rightmost subexpressions to be evaluated
into an accumulator register (eg, eax, rax, or r0) rather than onto
the stack.

Review URL: http://codereview.chromium.org/541047

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3630 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-01-18 14:04:55 +00:00
fschneider@chromium.org
637a21952f Added general pre- and postfix count operations to top-level compiler.
Until now we only supported postfix operations on global variables.
This change add generic count operations to the top-level compiler.

I tried to re-use code from the code generator used for assignment expressions
where possible.

Review URL: http://codereview.chromium.org/496009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3530 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2010-01-04 13:56:31 +00:00
fschneider@chromium.org
2c9b2ab78a Adding ThisFunction expression to top-level compiler.
Review URL: http://codereview.chromium.org/466056

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3430 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-12-07 15:10:39 +00:00
fschneider@chromium.org
3a21aafa7c Added fast compiler support for calling JS runtime functions.
Also added a simple test that invokes a JS runtime function
in top-level code.

Review URL: http://codereview.chromium.org/437081

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3371 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-27 00:28:06 +00:00
fschneider@chromium.org
1c90793581 Fix bug in the fast compiler's object literal code
Fixes issue 526:

BUG=http://code.google.com/p/v8/issues/detail?id=526

The object literals code in the fast compiler returned an incorrect result 
when getter or setters are defined together with computed properties.

Added a regression test that captures the most reduced version of this
problem. 

Also added a test for object literals with getters/setters and 
prototype properties.


Review URL: http://codereview.chromium.org/444001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3370 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-26 21:13:20 +00:00
kmillikin@chromium.org
1affb78938 Add support for for loops to the toplevel code generator.
Review URL: http://codereview.chromium.org/340059

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3199 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-02 14:12:48 +00:00
fschneider@chromium.org
e38cd233c3 Support for function calls on an arbitrary expression that returns
a function in the top-level compiler.

e.g.

function f() { return (function() { return true; }) }
f()()


Review URL: http://codereview.chromium.org/346029

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3196 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-02 12:04:35 +00:00
fschneider@chromium.org
a07146c0cb Support for post-fix count operations (x++, x--) where x is a global
variable for the top-level compiler.

Review URL: http://codereview.chromium.org/342058

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3194 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-02 10:22:22 +00:00
fschneider@chromium.org
07cd399b5c Support for calls on named and keyed properties in the fast compiler of the form:
o.x() and o[expr]()

other changes:
- Fix missing relocation info for StoreIC on global object.
- Generate only one common return sequence instead of always appending 
  "return <undefined>" at the end of each function: The first JS 
  return-statement will generate the common return sequence. All
  other return-statements will generate a unconditional branch to the common
  return sequence.


Review URL: http://codereview.chromium.org/340037

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3183 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-10-30 10:22:31 +00:00
fschneider@chromium.org
64e1d3205f Support for property access (named, keyed) in the fast compiler.
The generated code is similar to the existing code, but we never
inline any IC code in the fast compiler.


Review URL: http://codereview.chromium.org/337045

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3152 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-10-27 16:11:12 +00:00
fschneider@chromium.org
9445079c51 Support for object literals in fast compiler.
I also added more unit tests for literals.

Right now, the fast compiler produces code very similar to
the existing code generator. We may consider different ways to 
further compact the generated code for top-level code.

ARM always goes through a runtime function to initialize computed
properties in an object literal whereas IA32 and x64 use StoreIC.

Review URL: http://codereview.chromium.org/316009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3129 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-10-26 13:21:48 +00:00
kmillikin@chromium.org
fbc0eaa13d Added support for array literals to the toplevel compiler. They are
currently compiled the same as with the optimizing compiler: they are
cloned from a boilerplate object and the boilerplate objects are
lazily constructed.

Also changed argument pushing on ARM to use stm (store multiple),
which required changing the order of arguments to the runtime
functions DeclareGlobals and NewClosure.  They were only used from
generated code.

Finally, changed the toplevel code generator so that stack pops to
discard a temporary became addition to the stack pointer on ia32 and
x64.

Review URL: http://codereview.chromium.org/303021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3110 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-10-22 10:07:45 +00:00
kmillikin@chromium.org
61363d83ba Add support for global variable references in toplevel code. We use
the normal named load IC mechanism for now.  Generated code is similar
to the case for global variable assignments.

Review URL: http://codereview.chromium.org/294021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3101 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-10-21 09:38:21 +00:00
kmillikin@chromium.org
4dfd44479d Added support for assignments to global variables in the toplevel code
generator.  We use the normal store IC mechanism with the global
object as the receiver.  The following code is generated for 'x=true'
at toplevel.

======== IA32:

27  mov eax,0xf5d06161          ;; object: 0xf5d06161 <true>
32  mov ecx,0xf5d09c35          ;; object: 0xf5d09c35 <String[1]: x>
37  push [esi+0x17]
40  call StoreIC_Initialize  (0xf5ce75c0)    ;; code: STORE_IC, UNINITIALIZED
45  mov [esp],eax

======== X64:

25  movq rax,0x7f867a7b6199    ;; object: 0x7f867a7b6199 <true>
35  movq rcx,0x7f867a7bae71    ;; object: 0x7f867a7bae71 <String[1]: x>
45  push [rsi+0x2f]
49  call StoreIC_Initialize  (0x7f8655929ac0)    ;; code: STORE_IC, UNINITIALIZED
54  movq [rsp],rax

======== ARM:

32  e59f0054       ldr r0, [pc, #+84]          ;; object: 0xf5b78161 <true>
36  e59f2054       ldr r2, [pc, #+84]          ;; object: 0xf5b7bc35 <String[1]: x>
40  e598c017       ldr ip, [r8, #+23]
44  e52dc004       str ip, [sp, #-4]!
48  e1a0e00f       mov lr, pc
52  e59ff048       ldr pc, [pc, #+72]          ;; debug: statement 0
                                               ;; code: STORE_IC, UNINITIALIZED
56  e58d0000       str r0, [sp, #+0]

Review URL: http://codereview.chromium.org/305005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3095 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-10-20 13:37:26 +00:00
fschneider@chromium.org
d0dc31afab New mjsunit tests for the new compiler infrastructure. They
are in a separate subdirectory called 'compiler' in the mjsunit directory.

Review URL: http://codereview.chromium.org/273039

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3061 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-10-14 10:24:50 +00:00