ARM: Clean up literal pool generation....

ARM: Clean up literal pool generation.
Remove dead code, and generate pools less frequently.

BUG=none
TEST=none

Committed: http://code.google.com/p/v8/source/detail?r=8309

[Mads Ager (chromium), 2011-06-14 05:46:05]

Redirecting.

[Søren Thygesen Gjesse, 2011-06-14 07:50:56]

LGTM, but please consider the comments

And while you are changing these files how about renaming the prinfo in prinfo_
and num_prinfo_ to something more else e.g. pending_reloc_info?

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc
File src/arm/assembler-arm.cc (right):

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:327: first_const_pool_use_ = 0;
Maybe initialize to -1, as the value is only valid when num_prinfo_ is > 0.

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:2589: //    the constant pool is at least
kMaxDistToPool / 2.
And then add ASSERT(first_const_pool_use_ >= 0).

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:2628: ASSERT(IsLdrPcImmediateOffset(instr) && (instr &
kOff12Mask) == 0);
How about adding GetLdrPcImmediateOffsetOffset?

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:2637: instr_at_put(rinfo.pc(), instr + delta);
How about adding SetLdrPcImmediateOffsetOffset?

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:2642: first_const_pool_use_ = 0;
0 -> -1?

[Rodolph Perfetta, 2011-06-14 22:52:20]

Comments addressed, new patch will be uploaded soon.

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc
File src/arm/assembler-arm.cc (right):

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:327: first_const_pool_use_ = 0;
On 2011/06/14 07:50:56, Søren Gjesse wrote:
> Maybe initialize to -1, as the value is only valid when num_prinfo_ is > 0.

Done.

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:2589: //    the constant pool is at least
kMaxDistToPool / 2.
On 2011/06/14 07:50:56, Søren Gjesse wrote:
> And then add ASSERT(first_const_pool_use_ >= 0).

Done.

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:2628: ASSERT(IsLdrPcImmediateOffset(instr) && (instr &
kOff12Mask) == 0);
Used GetLdrRegImmediateOffset which was already implemented.

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:2637: instr_at_put(rinfo.pc(), instr + delta);
Used SetLdrRegImmediateOffset which was already implemented.

http://codereview.chromium.org/7108061/diff/4001/src/arm/assembler-arm.cc#new...
src/arm/assembler-arm.cc:2642: first_const_pool_use_ = 0;
On 2011/06/14 07:50:56, Søren Gjesse wrote:
> 0 -> -1?

Done.

[Søren Thygesen Gjesse, 2011-06-16 06:55:17]

LGTM

I will commit it.