Nop bytecodes are required only for break locations in debugger. Since nop bytecode doesn't change program state we can remove all of them.
There are at least two changes which this CL produce:
- we don't provide break position when we load local variable (still provide when load variable from global),
- we don't provide break position for statements without actual break positions (e.g. "a;") - these expressions should be super rare and user always can set breakpoint before or after this statement.
More details in one pager: https://docs.google.com/a/google.com/document/d/1JXlQpfMa9vRojbE272b6GMBbrfh6m_00135iAUOJEz8/edit?usp=sharing
Bug: v8:6425
Change-Id: I4aee73d497a84f7b5d89caa6dda6d3060567dfda
Reviewed-on: https://chromium-review.googlesource.com/543161
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Aleksey Kozyatinskiy <kozyatinskiy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46742}
Goal of this CL: explicit return from non-async function has position after
return expression as return position (will unblock [1]).
BytecodeArrayBuilder has SetStatementPosition and SetExpressionPosition methods.
If one of these methods is called then next generated bytecode will get passed
position. It's general treatment for most cases.
Unfortunately it doesn't work for Returns:
- debugger requires source positions exactly on kReturn bytecode in stepping
implementation,
- BytecodeGenerator::BuildReturn and BytecodeGenerator::BuildAsyncReturn
generates more then one bytecode and general solution will put return position
on first generated bytecode,
- it's not easy to split BuildReturn function into two parts to allow something
like following in BytecodeGenerator::VisitReturnStatement since generated
bytecodes are actually controlled by execution_control().
..->BuildReturnPrologue();
..->SetReturnPosition(stmt);
..->Return();
In this CL we pass ReturnStatement through ExecutionControl and use it for
position when we emit return bytecode right here.
So this CL only will improve return position for returns inside of non-async
functions, I'll address async functions later.
[1] https://chromium-review.googlesource.com/c/543161/
Change-Id: Iede512c120b00c209990bf50c20e7d23dc0d65db
Reviewed-on: https://chromium-review.googlesource.com/560738
Commit-Queue: Aleksey Kozyatinskiy <kozyatinskiy@chromium.org>
Reviewed-by: Adam Klein <adamk@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46687}
In current implementation in expressions like await foo() we have break location right after foo call and before actual await.
And we additionally have a lot of other statement locations because of do scope.
Let's move async debugging closer to sync debugging and introduce only one break location for await - before awaited function call.
Bug: v8:6425,v8:6162
Change-Id: I7568767856022c49101e7f3b7e39a2e401d21644
Reviewed-on: https://chromium-review.googlesource.com/514046
Reviewed-by: Marja Hölttä <marja@chromium.org>
Reviewed-by: Dmitry Gozman <dgozman@chromium.org>
Commit-Queue: Aleksey Kozyatinskiy <kozyatinskiy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45625}
There are two break locations at the same source location by desugaring:
- call iterator.next,
- before variable assignment.
Additionally location for for..of loops is moved from before "of" to before each variable expression.
We should not report first implicit call to avoid user confusion. User still able to go into .next function with both scenarios:
- when this call is reached by stepOver or stepInto from previous line,
- when this call is reached because of breakpoint at current line.
BUG=v8:6425
R=dgozman@chromium.org,jgruber@chromium.org
Review-Url: https://codereview.chromium.org/2893313002
Cr-Commit-Position: refs/heads/master@{#45509}
Split BytecodeGenerator::VisitSuspend into two pieces, one for
building the suspension code and one for resumption (these
are split into separate Build methods for convenience).
Each gets its own RegisterAllocationScope, which allows us to
reduce the register file size of the empty generator by 1.
For consistency, rename VisitGeneratorPrologue() to
BuildGeneratorPrologue() to match the names of the two
newly-created methods.
This relands the patch originally committed in
98927ea51b, as the test failure
due to that change was a code flushing bug. Code flushing was
disabled in de4a4095cf.
R=rmcilroy@chromium.org
Bug: v8:6379
Change-Id: Ifb4deafea99693c0a4e8646cf4e9884c7374cfc6
Reviewed-on: https://chromium-review.googlesource.com/508814
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Commit-Queue: Adam Klein <adamk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45406}
This reverts commit 98927ea51b.
Reason for revert: Breaks Mac GC Stress bot.
https://uberchromegw.corp.google.com/i/client.v8/builders/V8%20Mac%20GC%20Stress/builds/13299/steps/Mjsunit/logs/for-await-of
Original change's description:
> [ignition] Be smarter about register allocation in VisitSuspend
>
> Split BytecodeGenerator::VisitSuspend into two pieces, one for
> building the suspension code and one for resumption (these
> are split into separate Build methods for convenience).
> Each gets its own RegisterAllocationScope, which allows us to
> reduce the register file size of the empty generator by 1.
>
> For consistency, rename VisitGeneratorPrologue() to
> BuildGeneratorPrologue() to match the names of the two
> newly-created methods.
>
> Bug: v8:6379
> Change-Id: I08a617a44f99706cfff09bf86fb0a25a9cd6e032
> Reviewed-on: https://chromium-review.googlesource.com/503593
> Commit-Queue: Adam Klein <adamk@chromium.org>
> Reviewed-by: Aleksey Kozyatinskiy <kozyatinskiy@chromium.org>
> Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#45318}
TBR=rmcilroy@chromium.org,adamk@chromium.org,neis@chromium.org,kozyatinskiy@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Bug: v8:6379
Change-Id: I787fc3811c4f33a8021cf9170d43a74ed9b55d1c
Reviewed-on: https://chromium-review.googlesource.com/506548
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Commit-Queue: Deepti Gandluri <gdeepti@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45319}
Split BytecodeGenerator::VisitSuspend into two pieces, one for
building the suspension code and one for resumption (these
are split into separate Build methods for convenience).
Each gets its own RegisterAllocationScope, which allows us to
reduce the register file size of the empty generator by 1.
For consistency, rename VisitGeneratorPrologue() to
BuildGeneratorPrologue() to match the names of the two
newly-created methods.
Bug: v8:6379
Change-Id: I08a617a44f99706cfff09bf86fb0a25a9cd6e032
Reviewed-on: https://chromium-review.googlesource.com/503593
Commit-Queue: Adam Klein <adamk@chromium.org>
Reviewed-by: Aleksey Kozyatinskiy <kozyatinskiy@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45318}
Removes the do-expression wrapping, modifies BytecodeGenerator change
to enter a class literal's block scope if needed.
This does not solve the actual bug in v8:6322, but helps mitigate it in
simple cases. The bug is caused by BytecodeGenerator not allocating a
large enough array of context registers to hold its entire stack,
allowing non-context registers to be overwritten during PushContext and
PopContext bytecodes.
Nevertheless, I like the idea of not depending on do-expressions when
possible, so I think it's worth doing anyways.
BUG=v8:6322
R=rmcilroy@chromium.org, marja@chromium.org, littledan@chromium.org
Change-Id: I82b7569db2a0eead1694bd04765fc4456c2f1a0a
Reviewed-on: https://chromium-review.googlesource.com/491074
Commit-Queue: Caitlin Potter <caitp@igalia.com>
Reviewed-by: Marja Hölttä <marja@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Daniel Ehrenberg <littledan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45110}
Indisputable profit:
- correct break location in next task (see tests),
- stepOver with async await never lands in random code (see related test and issue),
- inspector doesn't store current stepping state in debugger agent and completely trust V8 - step to new inspector-V8 design (I will finish design doc soon).
- willExecuteScript and didExecuteScript instrumentation could be removed from code base - reduce probability of future errors.
- finally - less code,
- stepping implementation in V8 makes another step to follow our stepping strategy (stepOut should do stepInto and break when exit current frame) (another one one page design doc based on @aandrey comment is coming),
- knowledge about existing of context groups is still inspector-only.
Disputable part is related to super rare scenario when in single isolate we have more then one context group id with enabled debugger agent:
- if one agent request break in own context (stepping, pause, e.t.c.) then we ignore all breaks in another agent. From one hand it looks like good: user clicks stepInto and they don't expect that execution could be paused by another instance of DevTools in unobservable from current DevTools way (second DevTools will get paused notification and run nested message loop). From another hand we shouldn't ignore breakpoints or debugger statement never. In general, I think that proposed behavior is rathe feature then issue.
- and disadvantage, on attempt to break in non-target context group id we just call StepOut until reach target context group id, step out call could deoptimize code in non related to current debugger agent context. But break could happens only in case of debugger stmt or breakpoint - sound like minor issue. Ignoring break on exception sounds like real issue but by module of rareness of this case I think we can ignore this.
Implementation details:
- when debugger agent request break for any reason it passes target context group id to V8Debugger - last agent requesting break is preferred.
- when V8Debugger gets BreakProgramRequested notification from V8, it checks current context group id against target context group id, if they match then just process break as usual otherwise makes StepOut action,
- debug.cc at the end of microtask if last_scheduled_action is StepOut, schedules StepIn and will break on first instruction in next task.
BUG=chromium:654022
R=dgozman@chromium.org,yangguo@chromium.org
Review-Url: https://codereview.chromium.org/2748503002
Cr-Commit-Position: refs/heads/master@{#44034}
