Oilpan: Make Oilpan crash at a never-inlined function when hitting OOM

Oilpan: Make Oilpan crash at a never-inlined function when hitting OOM

In order to get reliable crash reports of OOM, we want to make Oilpan
crash at a never-inlined function. This CL adds blinkGCOutOfMemory for that purpose.

BUG=420515
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=192760

[haraken, 2015-03-30 07:38:54]

haraken@chromium.org changed reviewers:
+ oilpan-reviews@chromium.org, sigbjornf@opera.com

[haraken, 2015-03-30 07:38:55]

PTAL

[sof, 2015-03-30 08:06:33]

lgtm % issue below addressed.

https://codereview.chromium.org/1042263002/diff/1/Source/platform/heap/Heap.cpp
File Source/platform/heap/Heap.cpp (right):

https://codereview.chromium.org/1042263002/diff/1/Source/platform/heap/Heap.c...
Source/platform/heap/Heap.cpp:175: IMMEDIATE_CRASH();
Could you mimic PartitionAlloc's partitionOutOfMemoryWithLotsOfUncommitedPages()
? It has some Windows smarts.

[haraken, 2015-03-30 09:24:13]

https://codereview.chromium.org/1042263002/diff/1/Source/platform/heap/Heap.cpp
File Source/platform/heap/Heap.cpp (right):

https://codereview.chromium.org/1042263002/diff/1/Source/platform/heap/Heap.c...
Source/platform/heap/Heap.cpp:175: IMMEDIATE_CRASH();
On 2015/03/30 08:06:33, sof wrote:
> Could you mimic PartitionAlloc's
partitionOutOfMemoryWithLotsOfUncommitedPages()
> ? It has some Windows smarts.

I investigated but it seems it's involved (we need to implement
Heap::s_committedSpace and Heap::s_uncommittedSpace). May I add a TODO at the
moment and address it in a follow-up?

In case of PartitionAlloc, it aggressively allocates new virtual address spaces
(freed pages are never reused for different size buckets, and there are tons of
buckets in PartitionAlloc), so the virtual address space OOM is a concern. On
the other hand, I don't think it's a serious concern in Oilpan, intuitively.

[sof, 2015-03-30 13:37:49]

On 2015/03/30 09:24:13, haraken wrote:
>
https://codereview.chromium.org/1042263002/diff/1/Source/platform/heap/Heap.cpp
> File Source/platform/heap/Heap.cpp (right):
> 
>
https://codereview.chromium.org/1042263002/diff/1/Source/platform/heap/Heap.c...
> Source/platform/heap/Heap.cpp:175: IMMEDIATE_CRASH();
> On 2015/03/30 08:06:33, sof wrote:
> > Could you mimic PartitionAlloc's
> partitionOutOfMemoryWithLotsOfUncommitedPages()
> > ? It has some Windows smarts.
> 
> I investigated but it seems it's involved (we need to implement
> Heap::s_committedSpace and Heap::s_uncommittedSpace). May I add a TODO at the
> moment and address it in a follow-up?
> 

Certainly ok, I wasn't clear enough - I meant just use its (assumedly
hard-earned & valid) trick of crashing at a unique low address so as to make it
reliable on Windows / with MSVC also.

[haraken, 2015-03-30 13:48:43]

On 2015/03/30 13:37:49, sof wrote:
> On 2015/03/30 09:24:13, haraken wrote:
> >
>
https://codereview.chromium.org/1042263002/diff/1/Source/platform/heap/Heap.cpp
> > File Source/platform/heap/Heap.cpp (right):
> > 
> >
>
https://codereview.chromium.org/1042263002/diff/1/Source/platform/heap/Heap.c...
> > Source/platform/heap/Heap.cpp:175: IMMEDIATE_CRASH();
> > On 2015/03/30 08:06:33, sof wrote:
> > > Could you mimic PartitionAlloc's
> > partitionOutOfMemoryWithLotsOfUncommitedPages()
> > > ? It has some Windows smarts.
> > 
> > I investigated but it seems it's involved (we need to implement
> > Heap::s_committedSpace and Heap::s_uncommittedSpace). May I add a TODO at
the
> > moment and address it in a follow-up?
> > 
> 
> Certainly ok, I wasn't clear enough - I meant just use its (assumedly
> hard-earned & valid) trick of crashing at a unique low address so as to make
it
> reliable on Windows / with MSVC also.

Ah, makes sense; done.

[haraken, 2015-03-30 13:48:46]

The CQ bit was checked by haraken@chromium.org

[haraken, 2015-03-30 13:48:47]

The patchset sent to the CQ was uploaded after l-g-t-m from sigbjornf@opera.com
Link to the patchset: https://codereview.chromium.org/1042263002/#ps40001
(title: " ")

[commit-bot: I haz the power, 2015-03-30 13:48:57]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1042263002/40001

[haraken, 2015-03-30 13:49:30]

The CQ bit was checked by haraken@chromium.org

[haraken, 2015-03-30 13:49:30]

The patchset sent to the CQ was uploaded after l-g-t-m from sigbjornf@opera.com
Link to the patchset: https://codereview.chromium.org/1042263002/#ps60001
(title: " ")

[commit-bot: I haz the power, 2015-03-30 13:49:38]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1042263002/60001

[commit-bot: I haz the power, 2015-03-30 15:02:00]

Committed patchset #4 (id:60001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=192760