Get rid of the tryFast*alloc API.

Get rid of the tryFast*alloc API.

Also removes WTF_MALLOC_VALIDATION, which we do not use and therefore the complexity it adds is unwelcome.

There are various benefits:
1) The standard ones from having less code and less complexity.
2) The API doesn't actually work, so every new caller becomes a possible bug. On (at least) Linux and Windows, our underlying malloc is tcmalloc which is hard-wired to crash on allocation failure. So tryFast*alloc can never return a fail for most of our users.
3) We've removed an inconsistency between platforms: I think tryFast*alloc might actually work as advertised on Mac, because of the built-in copy of tcmalloc.

This also removes 500KB of guff from the Linux x64 chrome debug binary. A negligible effect on the release binary, sadly. 


BUG=none
TEST=compile, webkit_tests pass

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=155495

[abarth-chromium, 2013-08-04 19:21:31]

LGTM

I'm not 100% sure about changing CRASH to IMMEDIATE_CRASH.  Maybe that should be
a separate CL?

https://codereview.chromium.org/22038002/diff/4001/Source/wtf/FastMalloc.cpp
File Source/wtf/FastMalloc.cpp (right):

https://codereview.chromium.org/22038002/diff/4001/Source/wtf/FastMalloc.cpp#...
Source/wtf/FastMalloc.cpp:3385: IMMEDIATE_CRASH();
How is IMMEDIATE_CRASH different from CRASH?

[Chris Evans, 2013-08-04 20:31:27]

Sounds reasonable. There's another minor optimization we could make in the same
function so I'll spin off another CL if I decide to do them.

https://codereview.chromium.org/22038002/diff/4001/Source/wtf/FastMalloc.cpp
File Source/wtf/FastMalloc.cpp (right):

https://codereview.chromium.org/22038002/diff/4001/Source/wtf/FastMalloc.cpp#...
Source/wtf/FastMalloc.cpp:3385: IMMEDIATE_CRASH();
On 2013/08/04 19:21:31, abarth wrote:
> How is IMMEDIATE_CRASH different from CRASH?

It gets the job done with the minimum number of opcodes. In particular, it
doesn't dereference 0xbbadbeef before the final "guaranteed" crash instruction.

I don't think byte counting matters for this function, so I'll leave it alone to
land as you suggest.

[commit-bot: I haz the power, 2013-08-04 20:36:46]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/cevans@chromium.org/22038002/13001

[commit-bot: I haz the power, 2013-08-04 23:58:08]

Change committed as 155495