[wasm] WIP fix arm64 frame alignment.

[wasm] Fix arm64 frame alignment.

    In JavaScript code and stubs, JSSP mirrors the CSP but may be unaligned.
    But in WASM code only CSP is used, like native code, and it must be
    aligned.
    Calls into WASM from JS need to carefully align the C stack
    pointer (csp) and restore the previous JSSP, while calls from WASM
    to JS need to compute a new JSSP and restore their CSP after the
    call.

R=ahaas@chromium.org
BUG=

Committed: https://crrev.com/c9f59780f7f2429ad1dd6b870d4e7764747853a3
Cr-Commit-Position: refs/heads/master@{#35096}

[titzer, 2016-03-18 10:40:32]

Description was changed from

==========
[wasm] WIP fix arm64 frame alignment.

R=ahaas@chromium.org
BUG=
==========

to

==========
[wasm] Fix arm64 frame alignment.
    
    In JavaScript code and stubs, JSSP mirrors the CSP but may be unaligned.
    But in WASM code only CSP is used, like native code, and it must be
    aligned.
    Calls into WASM from JS need to carefully align the C stack
    pointer (csp) and restore the previous JSSP, while calls from WASM
    to JS need to compute a new JSSP and restore their CSP after the
    call.



R=ahaas@chromium.org
BUG=
==========

[titzer, 2016-03-24 10:11:34]

titzer@chromium.org changed reviewers:
+ danno@chromium.org

[titzer, 2016-03-24 10:11:34]

+danno

PTAL. I'd like to land this while the arm64 folks take a look at one
--always-opt failure.

[ahaas, 2016-03-29 08:33:58]

https://codereview.chromium.org/1811283003/diff/60001/src/arm64/macro-assembl...
File src/arm64/macro-assembler-arm64.cc (right):

https://codereview.chromium.org/1811283003/diff/60001/src/arm64/macro-assembl...
src/arm64/macro-assembler-arm64.cc:1358: void MacroAssembler::AssertCspAligned()
{
Could you let this function return a boolean so that you can call it in a
DCHECK?

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/code...
File src/compiler/arm64/code-generator-arm64.cc (right):

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/code...
src/compiler/arm64/code-generator-arm64.cc:538:
static_cast<CallDescriptor::Flags>(MiscField::decode(opcode));
You could use "CallDescriptor::Flags flags(MiscField::decode(instr->opcode()));"
here.

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/code...
src/compiler/arm64/code-generator-arm64.cc:1000: if (count > 0) {
count is <= 0 here, so this condition is always false.

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/inst...
File src/compiler/arm64/instruction-selector-arm64.cc (right):

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/inst...
src/compiler/arm64/instruction-selector-arm64.cc:1647: Emit(claim, g.NoOutput(),
g.TempImmediate(claim_count));
Could you add a comment here why "claim(0)" isn't a nop?

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/instructio...
File src/compiler/instruction-selector.cc (right):

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/instructio...
src/compiler/instruction-selector.cc:1547: bool caller_native_stack =
Similar to the code in the instruction-selector-arm64 you could call this
variable "from_native_stack".

[titzer, 2016-03-29 09:10:53]

https://codereview.chromium.org/1811283003/diff/60001/src/arm64/macro-assembl...
File src/arm64/macro-assembler-arm64.cc (right):

https://codereview.chromium.org/1811283003/diff/60001/src/arm64/macro-assembl...
src/arm64/macro-assembler-arm64.cc:1358: void MacroAssembler::AssertCspAligned()
{
On 2016/03/29 08:33:58, ahaas wrote:
> Could you let this function return a boolean so that you can call it in a
> DCHECK?

This method doesn't assert anything in C++ land; it generates an access through
the stack pointer, which functions as an assertion in the generated code.

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/code...
File src/compiler/arm64/code-generator-arm64.cc (right):

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/code...
src/compiler/arm64/code-generator-arm64.cc:538:
static_cast<CallDescriptor::Flags>(MiscField::decode(opcode));
On 2016/03/29 08:33:58, ahaas wrote:
> You could use "CallDescriptor::Flags
flags(MiscField::decode(instr->opcode()));"
> here.

Done.

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/code...
src/compiler/arm64/code-generator-arm64.cc:1000: if (count > 0) {
On 2016/03/29 08:33:58, ahaas wrote:
> count is <= 0 here, so this condition is always false.

There's no need to rely on that invariant, since it is only established by the
tricky logic in the instruction selector.

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/inst...
File src/compiler/arm64/instruction-selector-arm64.cc (right):

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/arm64/inst...
src/compiler/arm64/instruction-selector-arm64.cc:1647: Emit(claim, g.NoOutput(),
g.TempImmediate(claim_count));
On 2016/03/29 08:33:58, ahaas wrote:
> Could you add a comment here why "claim(0)" isn't a nop?

Done.

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/instructio...
File src/compiler/instruction-selector.cc (right):

https://codereview.chromium.org/1811283003/diff/60001/src/compiler/instructio...
src/compiler/instruction-selector.cc:1547: bool caller_native_stack =
On 2016/03/29 08:33:58, ahaas wrote:
> Similar to the code in the instruction-selector-arm64 you could call this
> variable "from_native_stack".

Done.

[ahaas, 2016-03-29 09:22:45]

lgtm

[titzer, 2016-03-29 10:38:39]

The CQ bit was checked by titzer@chromium.org

[commit-bot: I haz the power, 2016-03-29 10:38:50]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1811283003/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1811283003/100001

[commit-bot: I haz the power, 2016-03-29 11:41:39]

Description was changed from

==========
[wasm] Fix arm64 frame alignment.
    
    In JavaScript code and stubs, JSSP mirrors the CSP but may be unaligned.
    But in WASM code only CSP is used, like native code, and it must be
    aligned.
    Calls into WASM from JS need to carefully align the C stack
    pointer (csp) and restore the previous JSSP, while calls from WASM
    to JS need to compute a new JSSP and restore their CSP after the
    call.



R=ahaas@chromium.org
BUG=
==========

to

==========
[wasm] Fix arm64 frame alignment.
    
    In JavaScript code and stubs, JSSP mirrors the CSP but may be unaligned.
    But in WASM code only CSP is used, like native code, and it must be
    aligned.
    Calls into WASM from JS need to carefully align the C stack
    pointer (csp) and restore the previous JSSP, while calls from WASM
    to JS need to compute a new JSSP and restore their CSP after the
    call.



R=ahaas@chromium.org
BUG=
==========

[commit-bot: I haz the power, 2016-03-29 11:41:39]

Committed patchset #6 (id:100001)

[commit-bot: I haz the power, 2016-03-29 11:42:54]

Description was changed from

==========
[wasm] Fix arm64 frame alignment.
    
    In JavaScript code and stubs, JSSP mirrors the CSP but may be unaligned.
    But in WASM code only CSP is used, like native code, and it must be
    aligned.
    Calls into WASM from JS need to carefully align the C stack
    pointer (csp) and restore the previous JSSP, while calls from WASM
    to JS need to compute a new JSSP and restore their CSP after the
    call.



R=ahaas@chromium.org
BUG=
==========

to

==========
[wasm] Fix arm64 frame alignment.

    In JavaScript code and stubs, JSSP mirrors the CSP but may be unaligned.
    But in WASM code only CSP is used, like native code, and it must be
    aligned.
    Calls into WASM from JS need to carefully align the C stack
    pointer (csp) and restore the previous JSSP, while calls from WASM
    to JS need to compute a new JSSP and restore their CSP after the
    call.

R=ahaas@chromium.org
BUG=

Committed: https://crrev.com/c9f59780f7f2429ad1dd6b870d4e7764747853a3
Cr-Commit-Position: refs/heads/master@{#35096}
==========

[commit-bot: I haz the power, 2016-03-29 11:42:55]

Patchset 6 (id:??) landed as
https://crrev.com/c9f59780f7f2429ad1dd6b870d4e7764747853a3
Cr-Commit-Position: refs/heads/master@{#35096}