Add a level of indirection to exception handler addresses.

Add a level of indirection to exception handler addresses.

To support deoptimization of exception handlers, the handler address in the
stack is converted to a pair of code object and an index into a separate
table of code offsets.  The index part is invariant under deoptimization.
The index is packed into the handler state field so that handler size does
not change.

R=vegorov@chromium.org,fschneider@chromium.org
BUG=
TEST=

Committed: http://code.google.com/p/v8/source/detail?r=9975

[Kevin Millikin (Chromium), 2011-11-10 12:00:07]

This looks like a big change by the number of files, but many of them were
touched in only one place.

[Vyacheslav Egorov (Chromium), 2011-11-10 19:01:04]

lgtm

http://codereview.chromium.org/8462010/diff/1/src/arm/macro-assembler-arm.cc
File src/arm/macro-assembler-arm.cc (right):

http://codereview.chromium.org/8462010/diff/1/src/arm/macro-assembler-arm.cc#...
src/arm/macro-assembler-arm.cc:1309: // Compute the handler entry address and
jump to it.  The handler table is
Seems this code can be shared between throw and throw uncatchable.

http://codereview.chromium.org/8462010/diff/1/src/ia32/macro-assembler-ia32.cc
File src/ia32/macro-assembler-ia32.cc (right):

http://codereview.chromium.org/8462010/diff/1/src/ia32/macro-assembler-ia32.c...
src/ia32/macro-assembler-ia32.cc:884: jmp(edi);
Try to share code with Throw().

http://codereview.chromium.org/8462010/diff/1/src/objects.h
File src/objects.h (right):

http://codereview.chromium.org/8462010/diff/1/src/objects.h#newcode4063
src/objects.h:4063: static const int kHandlerTableOffset = kRelocationInfoOffset
+ kPointerSize;
I think this makes Code's header 64 bytes (with 28 bytes of padding) on ia32.

http://codereview.chromium.org/8462010/diff/1/src/parser.cc
File src/parser.cc (right):

http://codereview.chromium.org/8462010/diff/1/src/parser.cc#newcode496
src/parser.cc:496: int NextHandlerIndex() { return next_handler_index_++; }
Maybe call it AllocateHandlerIndex()? 

I would not expect side-effects from function called NextHandlerIndex.

http://codereview.chromium.org/8462010/diff/1/src/x64/macro-assembler-x64.cc
File src/x64/macro-assembler-x64.cc (right):

http://codereview.chromium.org/8462010/diff/1/src/x64/macro-assembler-x64.cc#...
src/x64/macro-assembler-x64.cc:2577: jmp(rdi);
Try to share code with throw.

[fschneider, 2011-11-11 09:43:13]

lgtm

http://codereview.chromium.org/8462010/diff/1/src/arm/macro-assembler-arm.cc
File src/arm/macro-assembler-arm.cc (right):

http://codereview.chromium.org/8462010/diff/1/src/arm/macro-assembler-arm.cc#...
src/arm/macro-assembler-arm.cc:1267: // The exception is expected in eax.
Should be r0 for arm.

[Kevin Millikin (Chromium), 2011-11-11 11:22:50]

http://codereview.chromium.org/8462010/diff/1/src/arm/macro-assembler-arm.cc
File src/arm/macro-assembler-arm.cc (right):

http://codereview.chromium.org/8462010/diff/1/src/arm/macro-assembler-arm.cc#...
src/arm/macro-assembler-arm.cc:1267: // The exception is expected in eax.
On 2011/11/11 09:43:13, fschneider wrote:
> Should be r0 for arm.

Thanks.  Silly that it was right and I changed it :)

http://codereview.chromium.org/8462010/diff/1/src/arm/macro-assembler-arm.cc#...
src/arm/macro-assembler-arm.cc:1309: // Compute the handler entry address and
jump to it.  The handler table is
On 2011/11/10 19:01:04, Vyacheslav Egorov wrote:
> Seems this code can be shared between throw and throw uncatchable.

Yeah, good idea.  Done.

http://codereview.chromium.org/8462010/diff/1/src/objects.h
File src/objects.h (right):

http://codereview.chromium.org/8462010/diff/1/src/objects.h#newcode4063
src/objects.h:4063: static const int kHandlerTableOffset = kRelocationInfoOffset
+ kPointerSize;
On 2011/11/10 19:01:04, Vyacheslav Egorov wrote:
> I think this makes Code's header 64 bytes (with 28 bytes of padding) on ia32.

That's disappointing.  I guess we could later choose to move this into the
deoptimization data to save the padding at the cost of an extra indirection when
throwing.

http://codereview.chromium.org/8462010/diff/1/src/parser.cc
File src/parser.cc (right):

http://codereview.chromium.org/8462010/diff/1/src/parser.cc#newcode496
src/parser.cc:496: int NextHandlerIndex() { return next_handler_index_++; }
On 2011/11/10 19:01:04, Vyacheslav Egorov wrote:
> Maybe call it AllocateHandlerIndex()? 
> 
> I would not expect side-effects from function called NextHandlerIndex.

I can see that, but I was following the style of NextMaterializedBlahBlahBlah
just above.  And AllocateMaterializedLiteralIndex is an eyeful.  Since those
materialized literal lines are already 78 characters long, I'm a-gonna keep it
this way :)