Commit Graph

19 Commits

Author SHA1 Message Date
jacob.bramley@arm.com
19a04c3a69 A64: Use a scope utility to allocate scratch registers.
This replaces Tmp0() and Tmp1() with a more flexible scratch register
pool. A scope-based utility can temporarily acquire registers from this
pool as needed.

We no longer have to worry about whether to use Tmp0(), Tmp1() or
something else; the scope can just get the next available scratch
register.

BUG=
R=jochen@chromium.org, rmcilroy@chromium.org

Review URL: https://codereview.chromium.org/164793003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19768 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-10 16:25:15 +00:00
bmeurer@chromium.org
bf86e624d4 Reland "Handle non-power-of-2 divisors in division-like operations".
Fixed the flooring div bug and added a test case.

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/191293012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19749 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-10 10:39:17 +00:00
svenpanne@chromium.org
fa6d25a602 Revert "Handle non-power-of-2 divisors in division-like operations", "A64 tweaks for division-like operations." and "Windows build fix.".
This reverts commit 19719, 19720 and 19721 because
mozilla/ecma/Date/15.9.3.1-1 fails (in release mode only?).

TBR=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/189963005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19722 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-07 13:11:56 +00:00
svenpanne@chromium.org
94c450fcb9 Handle non-power-of-2 divisors in division-like operations
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/190383002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19719 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-07 11:44:04 +00:00
alexandre.rames@arm.com
83c7b60293 A64: Move veneer emission checking in the Assembler.
The previous heuristic would break as a significant amount of code could be
generated without checking for veneer emission.
The veneer emission is now done in the Assembler, in a very similar way to
constant pool emission.

BUG=v8:3177
LOG=N
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/181873002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19661 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-04 15:54:12 +00:00
mvstanton@chromium.org
df988c7f6b The Array function must be looked up in the native context.
Platforms x64, a64, arm and mips had the bug that the array function was looked
up in the global context instead of the native context. Fix this, restoring a
weakened assert in hydrogen along the way (by the fix for crbug 347528, which
helped find this case).

R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/184383003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19597 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-28 10:39:36 +00:00
m.m.capewell@googlemail.com
0468660b13 A64: Tidy up register use in TaggedToI
Fix bug where input register was potentially corrupted, tidy up register use in
TruncateDoubleToI and rename TryInlineTruncateDoubleToI.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/173663002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19527 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-21 11:36:04 +00:00
alexandre.rames@arm.com
609af84e2b A64: Increase the margin for veneer checks.
BUG=v8:3168
LOG=N
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/171603008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19525 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-21 09:32:57 +00:00
jacob.bramley@arm.com
baf2614853 Remove the unused LoadInitialArrayMap.
BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/173883003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19513 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-20 16:36:53 +00:00
alexandre.rames@arm.com
7d64aab1a1 A64: Introduce 'branch types' that extend the architectural conditions.
The branch types include 'always' and 'never', and types like reg_zero (CBZ) and
reg_bit_clear (TBZ).
This will be used by incoming improvements to the code generated for
deoptimization exit points.

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/170783002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19497 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-19 16:13:09 +00:00
jochen@chromium.org
bbd233fc4a A64: Fix compilation with clang
The size_ member is only used in debug mode, so we should only define
it in debug mode.

BUG=none
R=rodolph.perfetta@arm.com, ulan@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/171583004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19495 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-19 15:20:15 +00:00
jacob.bramley@arm.com
d9b31caab3 A64: Fix r19474: Tidy up Push and Pop TODOs.
This fixes some snapshot=off, debug-mode regressions.

  - Call CheckConstPool when entering InsturctionAccurateScopes. This
    fixes a problem in PushMultipleTimes where 'count' is very high.
    Each call to PushHelper starts an InstructionAccurateScope
    immediately after the last one finishes, with no opportunity for the
    constant pool to be generated.

  - Temporarily disable AssertStackConsistency in PrepareForPush. This
    assertion seems to push the code size too far, and several cctests
    (intermittently) fail due to buffer size limitations. I would like
    to renable this assertion later.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/172453002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19491 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-19 14:37:18 +00:00
jacob.bramley@arm.com
6d9fcf1198 A64: Tidy up Push and Pop TODOs.
This addresses several TODOs:
  - Push and Pop requests can be queued up so that arrays of Registers
    can be pushed efficiently, with just one PrepareForPush/Pop.
  - PushMultipleTimes now takes an Operand. This allows variable-length
    arguments arrays to be initialized, for example.
  - A NoUseRealAbortsScope has been added to Abort so that
    AssertStackConsistency can be called from PrepareForPush without
    introducing infinite recursion.

BUG=
R=rmcilroy@chromium.org, ulan@chromium.org

Review URL: https://codereview.chromium.org/170623002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19474 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-19 09:43:45 +00:00
baptiste.afsa@arm.com
78b2a18d02 A64: Remove early_exit unused mechanism in CompareMap.
R=jochen@chromium.org, ulan@chromium.org

Review URL: https://codereview.chromium.org/169733003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19473 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-19 09:38:45 +00:00
alexandre.rames@arm.com
62116e2c12 A64: Let the MacroAssembler resolve branches to distant targets.
Code generation would fail when assembling a branch to a label that is bound
outside the immediate range of the instruction. A64 is sensitive to this, as the
various branching instructions have different ranges, going down to +-32KB for
TBZ/TBNZ.  The MacroAssembler is augmented to handle branches to targets that
may exceed the immediate range of instructions.

When branching backward to a label exceeding the instruction range, the
MacroAssembler can simply tweak the generated code to use an unconditional
branch with a longer range. For example instead of
    B(cond, &label);
the MacroAssembler can generate:
    b(InvertCondition(cond), &done);
    b(&label);
    bind(&done);

Since the target is not known when the branch is emitted, forward branches uses
a different mechanism. The MacroAssembler keeps track of forward branches to
unbound labels. When the code generation approaches the end of the range of a
branch, a veneer is generated for the branch.

BUG=v8:3148
LOG=Y
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/169893002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19444 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-18 13:15:32 +00:00
danno@chromium.org
bedf702bcb Revert r19403: "A64: Tidy up Push and Pop TODOs."
Causes a64 debug asserts

TBR=jacob.bramley@arm.com,ulan@chromium.org

Review URL: https://codereview.chromium.org/169303007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19419 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-17 16:08:44 +00:00
rmcilroy@chromium.org
f6e95dc928 A64 support for DoubleToIStub (truncating).
Added support for truncating DoubleToIStub and reorganize the macro-assembler
dToI operations to do the fast-path inline and the slow path by calling the
stub.

This a port essentially a port of https://codereview.chromium.org/23129003/.

R=jacob.bramley@arm.com, ulan@chromium.org

Review URL: https://codereview.chromium.org/160423002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19414 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-17 15:09:46 +00:00
jacob.bramley@arm.com
1634631ae4 A64: Tidy up Push and Pop TODOs.
This addresses several TODOs:
  - Push and Pop requests can be queued up so that arrays of Registers
    can be pushed efficiently, with just one PrepareForPush/Pop.
  - PushMultipleTimes now takes an Operand. This allows variable-length
    arguments arrays to be initialized, for example.
  - A NoUseRealAbortsScope has been added to Abort so that
    AssertStackConsistency can be called from PrepareForPush without
    introducing infinite recursion.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/169533002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19403 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-17 12:08:06 +00:00
ulan@chromium.org
e95bc7eec8 Merge experimental/a64 to bleeding_edge.
BUG=v8:3113
LOG=Y
R=jochen@chromium.org, rmcilroy@chromium.org, rodolph.perfetta@arm.com

Review URL: https://codereview.chromium.org/148293020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19311 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-02-12 09:19:30 +00:00