Change the handling of the debug break stack guard

Change the handling of the debug break stack guard. The debug break is no longer ignored when hit inside "system" JavaScript. The reason for this is twofold:

* Running "system" JavaScript with the debug break flag active leads to slow running code while waiting for the break in non "system" JavaScript (one exception to this it is to try to avoid breaks in the clear mirror cache JavaScript code called when leaving the debugger).

* If this happens while processing RegExp running in native code an infinite loop is created as the stack guard handler for RegExp does not move execution forward

Fixed a GC bug in the interrupt handling for RegExp running in native code.

Added test of debug break while in debug message handler callback and debug break while executing a RegExp.

Committed: http://code.google.com/p/v8/source/detail?r=2074

[Søren Thygesen Gjesse, 2009-05-28 14:46:55]

I have now revisited this change and it is ready for review.

[Erik Corry, 2009-05-29 08:10:28]

LGTM

http://codereview.chromium.org/115262/diff/22/24
File src/debug.h (right):

http://codereview.chromium.org/115262/diff/22/24#newcode291
Line 291: static void set_interrupt_pending(InterruptFlag what) {
Plural-singular discrepancy between comment and function name.

http://codereview.chromium.org/115262/diff/22/24#newcode690
Line 690: ASSERT(prev_ == NULL ? !Debug::is_interrupt_pending(PREEMPT) : true);
How about ASSERT(prev_ != NULL || !Debug::is_interrupt_pending(PREEMPT))

http://codereview.chromium.org/115262/diff/22/24#newcode728
Line 728: // Try to avoid any pending debug break to break in the clear mirror
to break -> breaking.
I guess this can't completely prevent breaks in the clear mirror code, so I
assume that breaks in the clear mirror code are not a correctness problem, just
a little confusing for the user?

http://codereview.chromium.org/115262/diff/22/24#newcode743
Line 743: Debug::clear_interrupt_pending(PREEMPT);
Should these two last lines be reordered?

[Søren Thygesen Gjesse, 2009-05-29 08:40:42]

http://codereview.chromium.org/115262/diff/22/24
File src/debug.h (right):

http://codereview.chromium.org/115262/diff/22/24#newcode291
Line 291: static void set_interrupt_pending(InterruptFlag what) {
On 2009/05/29 08:10:28, Erik Corry wrote:
> Plural-singular discrepancy between comment and function name.

Done.

http://codereview.chromium.org/115262/diff/22/24#newcode690
Line 690: ASSERT(prev_ == NULL ? !Debug::is_interrupt_pending(PREEMPT) : true);
On 2009/05/29 08:10:28, Erik Corry wrote:
> How about ASSERT(prev_ != NULL || !Debug::is_interrupt_pending(PREEMPT))

Done.

http://codereview.chromium.org/115262/diff/22/24#newcode728
Line 728: // Try to avoid any pending debug break to break in the clear mirror
On 2009/05/29 08:10:28, Erik Corry wrote:
> to break -> breaking.
> I guess this can't completely prevent breaks in the clear mirror code, so I
> assume that breaks in the clear mirror code are not a correctness problem,
just
> a little confusing for the user?

Done.

http://codereview.chromium.org/115262/diff/22/24#newcode743
Line 743: Debug::clear_interrupt_pending(PREEMPT);
On 2009/05/29 08:10:28, Erik Corry wrote:
> Should these two last lines be reordered?

Done.