Replace qsort with std::sort.

Replace qsort with std::sort.

std::sort is a template, so it can be inlined more aggressively. Also, it's
O(n log n), while libc's qsort is O(n^2)

BUG=2639
Committed: http://code.google.com/p/v8/source/detail?r=14406

[danno, 2013-04-23 07:27:13]

https://codereview.chromium.org/14315005/diff/6001/src/store-buffer.cc
File src/store-buffer.cc (right):

https://codereview.chromium.org/14315005/diff/6001/src/store-buffer.cc#newcod...
src/store-buffer.cc:261: std::sort(old_start_, old_top_);
Although I think I convinced myself that this is safe without the custom
CompareAddress function, wouldn't a safer change be to continue to specify the
custom comparison routine to std::sort?

https://codereview.chromium.org/14315005/diff/6001/src/utils.h
File src/utils.h (right):

https://codereview.chromium.org/14315005/diff/6001/src/utils.h#newcode456
src/utils.h:456: explicit RawComparer(int (*cmp)(const T*, const T*)) : cmp(cmp)
{}
can you use different names for the parameter and member variables please?

[jochen (gone - plz use gerrit), 2013-04-23 07:34:50]

https://codereview.chromium.org/14315005/diff/6001/src/store-buffer.cc
File src/store-buffer.cc (right):

https://codereview.chromium.org/14315005/diff/6001/src/store-buffer.cc#newcod...
src/store-buffer.cc:261: std::sort(old_start_, old_top_);
On 2013/04/23 07:27:13, danno wrote:
> Although I think I convinced myself that this is safe without the custom
> CompareAddress function, wouldn't a safer change be to continue to specify the
> custom comparison routine to std::sort?

Not sure why this should be safer? Almost all of the complexity of
CompareAddresses is because it tries to classify the pointers into smaller,
equal, and larger.

However, for sorting, a strict ordering (a less than b) is enough, and so we
don't need to worry about converting pointers to integers and not overflowing
them

https://codereview.chromium.org/14315005/diff/6001/src/utils.h
File src/utils.h (right):

https://codereview.chromium.org/14315005/diff/6001/src/utils.h#newcode456
src/utils.h:456: explicit RawComparer(int (*cmp)(const T*, const T*)) : cmp(cmp)
{}
On 2013/04/23 07:27:13, danno wrote:
> can you use different names for the parameter and member variables please?

Done.

[marja, 2013-04-23 09:08:05]

We shouldn't pass a custom comparison function unless there is a good reason for
doing so. It will prevent inlining the comparison and that'll be more
inefficient.

[marja, 2013-04-23 09:09:18]

Ehmh, I was being unclear. I meant: at least we shouldn't make the code so that
a custom comparison function will always be passed.

[danno, 2013-04-24 09:30:43]

lgtm, I'll land this for you

[danno, 2013-04-24 11:20:24]

Committed patchset #5 manually as r14406 (presubmit successful).