Fix the thresholds so that the heap does not grow uncontrollably. This fixes

Fix the thresholds so that the heap does not grow uncontrollably.  This fixes
the OOM on the 524 test, alllowing the underlying crash issue to become visible
again.
Committed: http://code.google.com/p/v8/source/detail?r=8918

[Vyacheslav Egorov (Chromium), 2011-08-11 15:11:34]

LGTM with comments.

http://codereview.chromium.org/7621014/diff/1/src/heap.cc
File src/heap.cc (right):

http://codereview.chromium.org/7621014/diff/1/src/heap.cc#newcode733
src/heap.cc:733: new_space()->Capacity() + AllocationFactor(old_gen_size));
maybe instead of XFactor you should call it XDelta. XIncrease, XGrowth, etc.

http://codereview.chromium.org/7621014/diff/1/src/heap.h
File src/heap.h (right):

http://codereview.chromium.org/7621014/diff/1/src/heap.h#newcode1216
src/heap.h:1216: static inline intptr_t PromotionFactor(intptr_t size) {
More comments about this magical functions would be appreciated.

http://codereview.chromium.org/7621014/diff/1/src/heap.h#newcode1226
src/heap.h:1226: PromotionFactor(bytes) * old_gen_limit_factor_,
multiplying factor by factor. confusing. more comments please.

old_gen_limit_factor_ is confusing name.

http://codereview.chromium.org/7621014/diff/1/src/spaces.cc
File src/spaces.cc (right):

http://codereview.chromium.org/7621014/diff/1/src/spaces.cc#newcode1933
src/spaces.cc:1933: if (FLAG_trace_gc && first_unswept_page_ != NULL) {
Ditto

http://codereview.chromium.org/7621014/diff/1/src/spaces.h
File src/spaces.h (right):

http://codereview.chromium.org/7621014/diff/1/src/spaces.h#newcode1448
src/spaces.h:1448: if (FLAG_trace_gc) {
I think it should be under verbose flag.

I prefer to keep trace_gc output clean and simple.

[Erik Corry, 2011-08-11 16:09:08]

I rewrote it because it was so confusing.

http://codereview.chromium.org/7621014/diff/1/src/heap.cc
File src/heap.cc (right):

http://codereview.chromium.org/7621014/diff/1/src/heap.cc#newcode733
src/heap.cc:733: new_space()->Capacity() + AllocationFactor(old_gen_size));
On 2011/08/11 15:11:34, Vyacheslav Egorov wrote:
> maybe instead of XFactor you should call it XDelta. XIncrease, XGrowth, etc.

This function does not exist any more.

http://codereview.chromium.org/7621014/diff/1/src/heap.h
File src/heap.h (right):

http://codereview.chromium.org/7621014/diff/1/src/heap.h#newcode1216
src/heap.h:1216: static inline intptr_t PromotionFactor(intptr_t size) {
On 2011/08/11 15:11:34, Vyacheslav Egorov wrote:
> More comments about this magical functions would be appreciated.

This function does not exist any more.

http://codereview.chromium.org/7621014/diff/1/src/heap.h#newcode1226
src/heap.h:1226: PromotionFactor(bytes) * old_gen_limit_factor_,
On 2011/08/11 15:11:34, Vyacheslav Egorov wrote:
> multiplying factor by factor. confusing. more comments please.
> 
> old_gen_limit_factor_ is confusing name. 

I was unable to find a better name, but I clarified the places it is used.

http://codereview.chromium.org/7621014/diff/1/src/spaces.h
File src/spaces.h (right):

http://codereview.chromium.org/7621014/diff/1/src/spaces.h#newcode1448
src/spaces.h:1448: if (FLAG_trace_gc) {
On 2011/08/11 15:11:34, Vyacheslav Egorov wrote:
> I think it should be under verbose flag.
> 
> I prefer to keep trace_gc output clean and simple.

Done.

[Vyacheslav Egorov (Chromium), 2011-08-12 09:11:51]

LGTM!

http://codereview.chromium.org/7621014/diff/5001/src/heap.cc
File src/heap.cc (right):

http://codereview.chromium.org/7621014/diff/5001/src/heap.cc#newcode738
src/heap.cc:738: old_gen_limit_factor_ = 1;
I think you wanted 2 here.

http://codereview.chromium.org/7621014/diff/5001/src/heap.h
File src/heap.h (right):

http://codereview.chromium.org/7621014/diff/5001/src/heap.h#newcode1233
src/heap.h:1233: limit += new_space_.Capacity();
new_space_.Capacity can change while lazy sweeping is in progress. 

So Lower... will fluctuate. Is it intentional?

http://codereview.chromium.org/7621014/diff/5001/src/spaces.cc
File src/spaces.cc (right):

http://codereview.chromium.org/7621014/diff/5001/src/spaces.cc#newcode1933
src/spaces.cc:1933: if (FLAG_trace_gc && first_unswept_page_ != NULL) {
_verbose

[Erik Corry, 2011-08-12 09:21:47]

http://codereview.chromium.org/7621014/diff/5001/src/heap.cc
File src/heap.cc (right):

http://codereview.chromium.org/7621014/diff/5001/src/heap.cc#newcode738
src/heap.cc:738: old_gen_limit_factor_ = 1;
On 2011/08/12 09:11:51, Vyacheslav Egorov wrote:
> I think you wanted 2 here.

Good point.

http://codereview.chromium.org/7621014/diff/5001/src/heap.h
File src/heap.h (right):

http://codereview.chromium.org/7621014/diff/5001/src/heap.h#newcode1233
src/heap.h:1233: limit += new_space_.Capacity();
On 2011/08/12 09:11:51, Vyacheslav Egorov wrote:
> new_space_.Capacity can change while lazy sweeping is in progress. 
> 
> So Lower... will fluctuate. Is it intentional?

No, but I don't think it matters.

http://codereview.chromium.org/7621014/diff/5001/src/spaces.cc
File src/spaces.cc (right):

http://codereview.chromium.org/7621014/diff/5001/src/spaces.cc#newcode1933
src/spaces.cc:1933: if (FLAG_trace_gc && first_unswept_page_ != NULL) {
On 2011/08/12 09:11:51, Vyacheslav Egorov wrote:
> _verbose

Actually this happens rarely and if it happens it might be an issue to so I will
leave it in.