kmillikin@chromium.org
6ed860b7d6
Remove branching and binding with two arguments from the JumpTarget
...
class. Manually manage off-frame results at the two sites where they
were used.
Review URL: http://codereview.chromium.org/125048
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2157 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-12 14:43:27 +00:00
kmillikin@chromium.org
b318835cdc
First round of JumpTarget simplification. Remove dead functions (jumps,
...
branches, and bind with more than two arguments). Remove unneeded
variable (the "watermark" on virtual frames).
Review URL: http://codereview.chromium.org/125045
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2153 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-12 11:25:13 +00:00
ager@chromium.org
f706cfa30c
Fix debugger after inlined keyed store change.
...
Make sure that the IC is always hit when debugging and make sure to
restore the fast case when leaving the debugger.
Review URL: http://codereview.chromium.org/125044
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2152 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-12 11:24:13 +00:00
ager@chromium.org
a8507cb43d
Inline keyed stores if the code is in a loop and the key is likely to
...
be a smi.
The inlined version works for stores to JSArrays where the key is a
smi that is within bounds of the array and the value is either
constant or a smi so we can skip the write-barrier.
Review URL: http://codereview.chromium.org/122035
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2144 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-11 13:17:26 +00:00
kmillikin@chromium.org
2d7d89d831
Get rid of the notion of "parameter pointer" that no longer exists in
...
the implementation but still lives on in comments and identifier
names.
Where it was used, it has been changed to "caller's stack pointer".
Review URL: http://codereview.chromium.org/118500
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2137 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-10 15:08:25 +00:00
kmillikin@chromium.org
35b5aab3c7
Fix a bug when shifting left by zero. Ensure that the left operand is
...
writable (non-aliased) so it can be used for the result in the slow
case.
Review URL: http://codereview.chromium.org/118496
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2134 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-10 12:41:57 +00:00
erik.corry@gmail.com
8bd85d8c82
Add instanceof stub for ARM.
...
Review URL: http://codereview.chromium.org/119353
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2132 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-10 11:42:13 +00:00
erik.corry@gmail.com
b7d48f5807
A bunch of changes to speed up math on ARM.
...
* Identify heap numbers that contain non-Smi int32s and do bit
ops on them without calling the fp hardware or emulation.
* Identify results that are non-Smi int32s and write them into
heap numbers without calling the fp hardware or emulation.
* Do unary minus on heap numbers without going into the runtime
system.
* On add, sub and mul if we have both Smi and heapnumber inputs
to the same operation then convert the Smi to a double and do
the op without going into runtime system. This also applies
if we have two Smi inputs but the result is not Smi.
Review URL: http://codereview.chromium.org/119241
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2131 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-10 10:20:37 +00:00
kmillikin@chromium.org
94a122296d
Cleanup of ARM exception handlers. Remove the unused code and
...
parameter pointer slots. Change it so that the handler address no
longer points into the middle of the handler.
Review URL: http://codereview.chromium.org/119414
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2129 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-10 09:00:07 +00:00
kmillikin@chromium.org
f6aa7ce3ee
Change the IA32 stack handler layout so that the handler address does
...
not point into the middle of the handler. Also remove top-of-stack
caching from MacroAssembler::PushTryHandler.
Review URL: http://codereview.chromium.org/119358
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2128 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-10 08:01:30 +00:00
kmillikin@chromium.org
4a96feeef3
Change IA32's CodeGenerator::GenerateFastCharCodeAt to eagerly
...
allocate and spill registers, so that the register reference counts
and virtual frame are unchanged in the main body.
This eliminates a few sites of magic branching or binding of
JumpTarget with arguments.
Review URL: http://codereview.chromium.org/119302
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2126 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-09 07:44:09 +00:00
kmillikin@chromium.org
0bfd850a26
Simplify the IA32 exception handler block by removing the unused code
...
pointer and parameter pointer slots.
Review URL: http://codereview.chromium.org/118383
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2123 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-08 14:39:50 +00:00
kmillikin@chromium.org
bd82b972fc
Simplify the processing of deferred code in the code generator. Our
...
deferred code snippets are highly stylized. They always make a call
to a stub or the runtime and then return. This change takes advantage
of that.
Creating a deferred code object now captures a snapshot of the
registers in the virtual frame. The registers are automatically saved
on entry to the deferred code and restored on exit.
The clients of deferred code must ensure that there is no change to
the registers in the virtual frame (eg, by allocating which can cause
spilling) or to the stack pointer. That is currently the case.
As a separate change, I will add either code to verify this constraint
or else code to forbid any frame effect.
The deferred code itself does not use the virtual frame or register
allocator (or even the code generator). It is raw macro assembler
code.
Review URL: http://codereview.chromium.org/118226
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2112 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-05 14:16:32 +00:00
kmillikin@chromium.org
84ef2d3ace
Refactor IA32 shift operations to simplify moving the right operand
...
into the ecx register and to ensure that there is no frame effect
between the first entry to the deferred code and binding its exit.
Review URL: http://codereview.chromium.org/118157
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2096 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-03 12:20:56 +00:00
kmillikin@chromium.org
a41b41bf98
As a simplification, manually inline the function
...
DeferredInlineBinaryOperation::GenerateInlineCode and remove its
definition. It was only called from one site and was the only
deferred code object that was split that way into fast-case inline and
slow-case stub.
Review URL: http://codereview.chromium.org/119037
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2090 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-02 20:11:26 +00:00
kmillikin@chromium.org
c70a511b70
Factor out the code for emitting the IA32 binary operations div and
...
mod so they do not share code with the other binary operations. They
now preallocate their fixed registers (eax and edx). There is now no
frame effect between entries to the deferred call to the stub.
Review URL: http://codereview.chromium.org/118110
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2087 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-02 11:18:41 +00:00
mikhail.naganov@gmail.com
0fcedde224
Fix determining of JS lower stack bottom used in profiler's JS stack tracer to work with Chromium.
...
My assumption that log initialization happens somewhere near the stack's bottom is true for V8's sample shell but isn't true for Chromium, causing many otherwise valid stack addresses to be thrown out. The solution proposed is to save stack pointer value for the outermost JS function in ThreadLocalTop similar to c_entry_fp.
Implemented only for IA-32. Currently I'm not dealing with profiling on ARM and x86-64 anyway.
Review URL: http://codereview.chromium.org/112082
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2086 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-06-02 09:33:17 +00:00
kmillikin@chromium.org
3d15415414
Simplify deferred code by removing some unneeded or redundant stuff.
...
Review URL: http://codereview.chromium.org/113895
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2065 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-27 15:13:00 +00:00
kmillikin@chromium.org
9c829fafe9
Change the register allocator so that it no longer tracks references
...
to the platform-specific reserved registers. They are always in use
for their intended purpose, cannot appear in the virtual frame, and
can be freely used without allocation in the code generator.
Review URL: http://codereview.chromium.org/113837
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2061 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-27 07:53:47 +00:00
erik.corry@gmail.com
01f7c87f22
This patch much improves our tracking of whether function is
...
called from within a loop or not. In the past we lost the
information if a call site went megamorphic before a lazily
compiled callee was called for the first time. Now we track
that correctly (this is an issue that affects richards).
We still don't manage to track the in-loop state through a
constructor call, since constructor calls use LoadICs instead
of CallICs. This issue affects delta-blue. So in this patch
we assume that lazy compilations that don't happen through a
CallIC happen from inside a loop. I have an idea to fix this
but this patch is big enough already.
With our improved tracking of in-loop state I have switched
off the inlining of in-object loads for code that is not in
a loop. This benefits compile speed. One issue is that
eagerly compiled code now doesn't get the in-object loads
inlined. We need to eagerly compile less code to fix this.
Review URL: http://codereview.chromium.org/115744
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2046 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-25 18:29:02 +00:00
mikhail.naganov@gmail.com
30a0a7de43
Split nested namespaces declaration in two lines in accordance with C++ Style Guide.
...
This issue was raised by Brett Wilson while reviewing my changelist for readability. Craig Silverstein (one of C++ SG maintainers) confirmed that we should declare one namespace per line. Our way of namespaces closing seems not violating style guides (there is no clear agreement on it), so I left it intact.
Review URL: http://codereview.chromium.org/115756
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2038 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-25 10:05:56 +00:00
kmillikin@chromium.org
bcff796134
Size reduction of VirtualFrame objects. Remove the code generator and
...
macro assembler pointers and all derived state.
Review URL: http://codereview.chromium.org/115564
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@2011 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-20 11:14:18 +00:00
ager@chromium.org
ef85ba46b0
First round of size reduction for JumpTargets. Reduce their size by
...
two words: there is no reason to keep a pointer to the current code
generator and macro assembler in the JumpTarget.
Review URL: http://codereview.chromium.org/113458
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1987 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-18 12:26:59 +00:00
kmillikin@chromium.org
20b3a4deaa
Allow a few more (multiple-entry) deferred code snippets to take the
...
fast-merge path by manually giving them an expected entry frame.
Review URL: http://codereview.chromium.org/113518
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1984 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-18 07:57:29 +00:00
iposva@chromium.org
5b72959f3e
Remove space at end of line.
...
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1979 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-15 23:05:21 +00:00
iposva@chromium.org
b5c6dc35f9
- Fix numeric overflow handling when compiling count operations.
...
In certain situations the overflow checking corrupted the value
being calculated.
Review URL: http://codereview.chromium.org/115423
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1978 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-15 22:36:26 +00:00
kmillikin@chromium.org
dc9473d817
Try to preallocate the scratch registers used for inlined keyed loads
...
to allow fast entry to the deferred code if possible.
Review URL: http://codereview.chromium.org/113457
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1968 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-15 12:43:00 +00:00
ager@chromium.org
a3f30f5a3a
Clean up the Result class. Reduce the size of Result from four words
...
to one by keeping a stack of active code generators and by using
indirection to handles.
Mainly a cleanup. No visible performance impact.
Review URL: http://codereview.chromium.org/113455
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1965 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-15 11:09:51 +00:00
bak@chromium.org
9e02f380f0
Removed the delete calls to ZoneObjects.
...
Review URL: http://codereview.chromium.org/115348
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1948 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-14 11:36:10 +00:00
kmillikin@chromium.org
f48a6bf30f
Add more detailed timers of the various compilation passes. The
...
aggregate compilation time timer is the same as it was before.
Review URL: http://codereview.chromium.org/115344
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1946 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-14 10:29:48 +00:00
kmillikin@chromium.org
ebbaeb3655
Initial bypass of JumpTarget::ComputeEntryFrame for deferred code
...
entry labels where it is statically safe.
Review URL: http://codereview.chromium.org/115296
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1943 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-14 07:38:44 +00:00
kmillikin@chromium.org
18f69a7171
Fix for issue 341. In the stub for instanceof, we could try to read
...
an object's map before we were sure it was a heap object.
Review URL: http://codereview.chromium.org/115236
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1914 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-12 11:40:14 +00:00
mikhail.naganov@gmail.com
3e82fce446
Restore stack backtrace tests removed in revision 1785.
...
To re-enable tests, instead of compiled code patching, inlined code is used.
Inlined code is only installed in test.
Review URL: http://codereview.chromium.org/108015
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1892 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-05-07 09:24:43 +00:00
whesse@chromium.org
714e525b40
Improve register allocation of left shift operation. Add tests
...
for all shift operations.
Review URL: http://codereview.chromium.org/101016
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1825 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-04-30 09:21:50 +00:00
erik.corry@gmail.com
9a9e09ee73
Unbreak IA32 generated code coverage.
...
Review URL: http://codereview.chromium.org/100156
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1814 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-04-29 11:04:28 +00:00
kmillikin@chromium.org
d30bc90b9c
Inline the inobject property case for named property loads.
...
Review URL: http://codereview.chromium.org/99120
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1806 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-04-28 10:40:36 +00:00
lrn@chromium.org
a4d756a1c8
Move backend specific files to separate directories.
...
Move ia32 and arm specific files to subdirectories to make it easier to add more backends.
Review URL: http://codereview.chromium.org/92068
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1782 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-04-23 12:06:38 +00:00