Issue 1218493005: Debugger: use debug break slots instead of ICs (except for calls).

Issue 1218493005: Debugger: use debug break slots instead of ICs (except for calls). (Closed)

Created:
5 years, 5 months ago by Yang

Modified:
5 years, 5 months ago

Reviewers:
ulan

CC:
v8-dev

Base URL:
https://chromium.googlesource.com/v8/v8.git@master

Target Ref:
refs/pending/heads/master

Project:
v8

Visibility:
Public.

More Reviews

Description

Debugger: use debug break slots instead of ICs (except for calls). BUG=v8:3147, v8:4269 LOG=N Committed: https://crrev.com/a8a4c364c2ad4c1e5f4c6e1f1580d75ff1a35a4f Cr-Commit-Position: refs/heads/master@{#29487}

Patch Set 1 #

Patch Set 2 : mips ports #

Patch Set 3 : minor fixes #

Total comments: 12

Patch Set 4 : addressed comments #

Created: 5 years, 5 months ago

Download [raw] [tar.bz2]

	Unified diffs	Side-by-side diffs	Delta from patch set	Stats (+375 lines, -1240 lines)			Patch
M	src/arm/debug-arm.cc	View		1 chunk	+0 lines, -44 lines	0 comments	Download
M	src/arm/full-codegen-arm.cc	View	1 2 3	31 chunks	+46 lines, -66 lines	0 comments	Download
M	src/arm64/debug-arm64.cc	View		1 chunk	+0 lines, -44 lines	0 comments	Download
M	src/arm64/full-codegen-arm64.cc	View	1 2 3	32 chunks	+27 lines, -48 lines	0 comments	Download
M	src/assembler.h	View		1 chunk	+0 lines, -23 lines	0 comments	Download
M	src/assembler.cc	View		2 chunks	+4 lines, -6 lines	0 comments	Download
M	src/builtins.h	View		1 chunk	+0 lines, -10 lines	0 comments	Download
M	src/builtins.cc	View	1 2 3	1 chunk	+0 lines, -25 lines	0 comments	Download
M	src/codegen.h	View		1 chunk	+0 lines, -4 lines	0 comments	Download
M	src/codegen.cc	View		1 chunk	+0 lines, -14 lines	0 comments	Download
M	src/debug.h	View	1 2 3	1 chunk	+0 lines, -1 line	0 comments	Download
M	src/debug.cc	View	1 2 3	9 chunks	+30 lines, -89 lines	0 comments	Download
M	src/full-codegen.h	View	1 2 3	2 chunks	+15 lines, -32 lines	0 comments	Download
M	src/full-codegen.cc	View	1 2 3	13 chunks	+58 lines, -332 lines	0 comments	Download
M	src/ia32/debug-ia32.cc	View		2 chunks	+0 lines, -46 lines	0 comments	Download
M	src/ia32/full-codegen-ia32.cc	View	1 2 3	31 chunks	+49 lines, -70 lines	0 comments	Download
M	src/mips/debug-mips.cc	View	1	1 chunk	+0 lines, -43 lines	0 comments	Download
M	src/mips/full-codegen-mips.cc	View	1 2 3	32 chunks	+36 lines, -52 lines	0 comments	Download
M	src/mips64/debug-mips64.cc	View	1	1 chunk	+0 lines, -42 lines	0 comments	Download
M	src/mips64/full-codegen-mips64.cc	View	1 2 3	33 chunks	+38 lines, -52 lines	0 comments	Download
M	src/parser.cc	View	1 2	3 chunks	+4 lines, -2 lines	0 comments	Download
M	src/x64/debug-x64.cc	View		1 chunk	+0 lines, -44 lines	0 comments	Download
M	src/x64/full-codegen-x64.cc	View	1 2 3	32 chunks	+32 lines, -52 lines	0 comments	Download
M	test/cctest/test-debug.cc	View		5 chunks	+19 lines, -81 lines	0 comments	Download
M	test/mjsunit/debug-backtrace.js	View		2 chunks	+2 lines, -2 lines	0 comments	Download
M	test/mjsunit/debug-breakpoints.js	View		1 chunk	+0 lines, -1 line	0 comments	Download
M	test/mjsunit/debug-stepframe.js	View		3 chunks	+4 lines, -2 lines	0 comments	Download
M	test/mjsunit/es6/debug-step-into-constructor.js	View		7 chunks	+10 lines, -9 lines	0 comments	Download
M	test/mjsunit/es6/debug-stepnext-for.js	View	1 2	1 chunk	+1 line, -1 line	0 comments	Download
M	test/mjsunit/mjsunit.status	View		1 chunk	+0 lines, -3 lines	0 comments	Download

Messages

Total messages: 11 (2 generated)

Expand Messages | Collapse Messages | Show Generated Messages | Hide Generated Messages

Yang

Design doc here: https://docs.google.com/document/d/1UfpzkLuLqVy6QfaYYbhXeR0_mT53Qg7UXf3E-TvNVEo/edit#heading=h.o3qhv46e0ele

5 years, 5 months ago (2015-07-03 13:54:06 UTC) #2

ulan

Nice clean-up! https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-arm.cc File src/arm/full-codegen-arm.cc (right): https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-arm.cc#newcode2130 src/arm/full-codegen-arm.cc:2130: SetExpressionPosition(expr); Why we don't insert break here? ...

5 years, 5 months ago (2015-07-06 09:24:19 UTC) #3

Yang

Replies inline. https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-arm.cc File src/arm/full-codegen-arm.cc (right): https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-arm.cc#newcode2130 src/arm/full-codegen-arm.cc:2130: SetExpressionPosition(expr); On 2015/07/06 09:24:18, ulan wrote: > ...

5 years, 5 months ago (2015-07-06 10:05:08 UTC) #4

ulan

On 2015/07/06 10:05:08, Yang wrote: > Replies inline. > > https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-arm.cc > File src/arm/full-codegen-arm.cc (right): ...

5 years, 5 months ago (2015-07-06 10:18:02 UTC) #5

On 2015/07/06 10:05:08, Yang wrote:
> Replies inline.
> 
>
https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-ar...
> File src/arm/full-codegen-arm.cc (right):
> 
>
https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-ar...
> src/arm/full-codegen-arm.cc:2130: SetExpressionPosition(expr);
> On 2015/07/06 09:24:18, ulan wrote:
> > Why we don't insert break here?
> 
> Because EmitReturnSequence emits a return, which is considered a statement and
> can be used to set a break point.
> 
>
https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-ar...
> src/arm/full-codegen-arm.cc:2832: SetExpressionPosition(expr);
> On 2015/07/06 09:24:18, ulan wrote:
> > Why we don't insert break here?
> 
> Actually we don't need this at all. The caller already emits a position.
> 
> https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.cc
> File src/full-codegen.cc (right):
> 
>
https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.cc#new...
> src/full-codegen.cc:410: bool RecordPositions(MacroAssembler* masm, int pos,
> bool statement = true) {
> On 2015/07/06 09:24:18, ulan wrote:
> > Please use enum instead of bool and avoid implicit.
> 
> Done.
> 
>
https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.cc#new...
> src/full-codegen.cc:783: SetStatementPosition(stmt, SKIP_BREAK);
> On 2015/07/06 09:24:18, ulan wrote:
> > Please add a comment that the break is added below.
> 
> Done.
> 
>
https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.cc#new...
> src/full-codegen.cc:850: SetStatementPosition(stmt, SKIP_BREAK);
> On 2015/07/06 09:24:18, ulan wrote:
> > Please add a comment that the break is added below (here and in other
> functions
> > too).
> 
> Done.
> 
> https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.h
> File src/full-codegen.h (right):
> 
>
https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.h#newc...
> src/full-codegen.h:665: void SetStatementPosition(Statement* stmt,
> On 2015/07/06 09:24:19, ulan wrote:
> > Can we make the insert_break explicit in SetStatementPosition and
> > SetExpressionPosition?
> > 
> > Since they are using opposite defaults call-sites become confusing. Also
> > SetReturnPosition and SetFunctionPosition do not record slots at all, but
> > SetExpressionAsStatementPosition does.
> > 
> > SetExpressionAsStatementPosition should be named
> > SetExpressionAsStatementPositionAndInsertBreak.
> 
> The background here is that we generally break at statements, but not at
> expressions. Expression positions are mostly just used for stack traces, not
> break points.
> 
> That's why the default is that statement --> break, expression --> no break.
Not
> having the default would mean that I would have to add a lot of boilerplate at
> all call sites except for a few exceptions.
> 
> Can I simply add a comment here?

I see. Just comment would be ok.

Yang

On 2015/07/06 10:18:02, ulan wrote: > On 2015/07/06 10:05:08, Yang wrote: > > Replies inline. ...

5 years, 5 months ago (2015-07-06 10:21:23 UTC) #6

On 2015/07/06 10:18:02, ulan wrote:
> On 2015/07/06 10:05:08, Yang wrote:
> > Replies inline.
> > 
> >
>
https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-ar...
> > File src/arm/full-codegen-arm.cc (right):
> > 
> >
>
https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-ar...
> > src/arm/full-codegen-arm.cc:2130: SetExpressionPosition(expr);
> > On 2015/07/06 09:24:18, ulan wrote:
> > > Why we don't insert break here?
> > 
> > Because EmitReturnSequence emits a return, which is considered a statement
and
> > can be used to set a break point.
> > 
> >
>
https://codereview.chromium.org/1218493005/diff/40001/src/arm/full-codegen-ar...
> > src/arm/full-codegen-arm.cc:2832: SetExpressionPosition(expr);
> > On 2015/07/06 09:24:18, ulan wrote:
> > > Why we don't insert break here?
> > 
> > Actually we don't need this at all. The caller already emits a position.
> > 
> > https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.cc
> > File src/full-codegen.cc (right):
> > 
> >
>
https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.cc#new...
> > src/full-codegen.cc:410: bool RecordPositions(MacroAssembler* masm, int pos,
> > bool statement = true) {
> > On 2015/07/06 09:24:18, ulan wrote:
> > > Please use enum instead of bool and avoid implicit.
> > 
> > Done.
> > 
> >
>
https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.cc#new...
> > src/full-codegen.cc:783: SetStatementPosition(stmt, SKIP_BREAK);
> > On 2015/07/06 09:24:18, ulan wrote:
> > > Please add a comment that the break is added below.
> > 
> > Done.
> > 
> >
>
https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.cc#new...
> > src/full-codegen.cc:850: SetStatementPosition(stmt, SKIP_BREAK);
> > On 2015/07/06 09:24:18, ulan wrote:
> > > Please add a comment that the break is added below (here and in other
> > functions
> > > too).
> > 
> > Done.
> > 
> > https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.h
> > File src/full-codegen.h (right):
> > 
> >
>
https://codereview.chromium.org/1218493005/diff/40001/src/full-codegen.h#newc...
> > src/full-codegen.h:665: void SetStatementPosition(Statement* stmt,
> > On 2015/07/06 09:24:19, ulan wrote:
> > > Can we make the insert_break explicit in SetStatementPosition and
> > > SetExpressionPosition?
> > > 
> > > Since they are using opposite defaults call-sites become confusing. Also
> > > SetReturnPosition and SetFunctionPosition do not record slots at all, but
> > > SetExpressionAsStatementPosition does.
> > > 
> > > SetExpressionAsStatementPosition should be named
> > > SetExpressionAsStatementPositionAndInsertBreak.
> > 
> > The background here is that we generally break at statements, but not at
> > expressions. Expression positions are mostly just used for stack traces, not
> > break points.
> > 
> > That's why the default is that statement --> break, expression --> no break.
> Not
> > having the default would mean that I would have to add a lot of boilerplate
at
> > all call sites except for a few exceptions.
> > 
> > Can I simply add a comment here?
> 
> I see. Just comment would be ok.

Thanks. Is there anything left do to for me?

commit-bot: I haz the power

CQ is trying da patch. Follow status at https://chromium-cq-status.appspot.com/patch-status/1218493005/60001

5 years, 5 months ago (2015-07-06 10:30:51 UTC) #9

commit-bot: I haz the power

5 years, 5 months ago (2015-07-06 11:16:10 UTC) #11

Message was sent while issue was closed.

Patchset 4 (id:??) landed as
https://crrev.com/a8a4c364c2ad4c1e5f4c6e1f1580d75ff1a35a4f
Cr-Commit-Position: refs/heads/master@{#29487}

Expand Messages | Collapse Messages | Show Generated Messages | Hide Generated Messages