Move the ASan error info struct and function into their own file.

Move the ASan error info struct and function into their own file.

Sorry for the size of this CL... Most of this code has just been moved, here's more details for the file with a big positive diff:
- asan_error_info.[h/cc]: All the access/error information structure and functions have been moved here, there's no new code here, just some code shuffling.
- asan_error_info_unittest.cc: only AsanErrorInfoTest.ErrorInfoGetAsanBlockInfo is new, the rest is just some code shuffling.
- block_utils.cc: new file (but only 8 lines long !).
- block_utils_unittest.cc: Tests have been moved from asan_heap_unittest.cc
- unittest_util.h: I've moved some useful fixtures that were used in asan_heap.cc, no new code.
BUG=
R=chrisha@chromium.org

Committed: https://code.google.com/p/sawbuck/source/detail?r=2195

[Sébastien Marchand, 2014-07-24 22:36:19]

PTAL. Sorry for the size of this CL... Most of this code has just been moved,
here's more details for the file with a big positive diff:
- asan_error_info.[h/cc]: All the access/error information structure and
functions have been moved here, there's no new code here, just some code
shuffling.
- asan_error_info_unittest.cc: only AsanErrorInfoTest.ErrorInfoGetAsanBlockInfo
is new, the rest is just some code shuffling.
- block_utils.cc: new file (but only 8 lines long !).
- block_utils_unittest.cc: Tests have been moved from asan_heap_unittest.cc
- unittest_util.h: I've moved some useful fixtures that were used in
asan_heap.cc, no new code.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
File syzygy/agent/asan/asan_error_info_unittest.cc (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
syzygy/agent/asan/asan_error_info_unittest.cc:153: const size_t kAllocSize =
100;
This is a new test !

[chrisha, 2014-07-25 15:07:46]

Calling this asan_error_info.h/.cc is a little redundant. error_info would
suffice. (I've been slowly removing the asan_ prefixes, as ideally the file name
should be identical to the contents... for example, asan_heap should be
heap_proxy, asan_heap_checker should be heap_checker, etc)

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
File syzygy/agent/asan/asan_error_info.cc (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
syzygy/agent/asan/asan_error_info.cc:54: memcpy(dst,
::memcpy

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
syzygy/agent/asan/asan_error_info.cc:77: case HEAP_BUFFER_OVERFLOW:
We tend to bracket multi-line case blocks with {}:

case FOO: {
  ...
  ...
  break;
}

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
File syzygy/agent/asan/asan_error_info_unittest.cc (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
syzygy/agent/asan/asan_error_info_unittest.cc:17: #include "windows.h"
This should be a <> include.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/block_u...
File syzygy/agent/asan/block_utils.h (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/block_u...
syzygy/agent/asan/block_utils.h:50: // Check if a block is corrupt. This check
the block's metadata and its
Checks* if ...  This checks* the

Add a comment about this requiring the block redzones to be readable? (This is
something we know need to start worrying about as we add page protections...)

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
File syzygy/agent/asan/unittest_util.cc (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.cc:134: memset(buffer, 0, sizeof(buffer));
::memset, here and elsewhere.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.cc:196: }
ubernit: No need for the {} on these single line loops (here and elsewhere).

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.cc:230:
common::AlignUp(reinterpret_cast<size_t>(user_ptr)+user_alloc_size,
Indent +2

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.cc:233: pos < buffer_align_begin +
asan_alloc_size;
Align with 'const'

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
File syzygy/agent/asan/unittest_util.h (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.h:324: // A unittest fixture that initialize an
ASan heap that can be used in the
initializes*

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.h:357: // +--- 0 <= size < 2048 bytes---+---2048
bytes---+--2048 bytes--+
We should grow this to support an alignment of up to 4096, as that's what we'll
need to support the about-to-land page protection features.

[Sébastien Marchand, 2014-07-25 15:46:06]

Thanks, PTAnL.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
File syzygy/agent/asan/asan_error_info.cc (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
syzygy/agent/asan/asan_error_info.cc:54: memcpy(dst,
On 2014/07/25 15:07:45, chrisha wrote:
> ::memcpy

Done.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
syzygy/agent/asan/asan_error_info.cc:77: case HEAP_BUFFER_OVERFLOW:
On 2014/07/25 15:07:45, chrisha wrote:
> We tend to bracket multi-line case blocks with {}:
> 
> case FOO: {
>   ...
>   ...
>   break;
> }

Done.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
File syzygy/agent/asan/asan_error_info_unittest.cc (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/asan_er...
syzygy/agent/asan/asan_error_info_unittest.cc:17: #include "windows.h"
On 2014/07/25 15:07:45, chrisha wrote:
> This should be a <> include.

Done.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/block_u...
File syzygy/agent/asan/block_utils.h (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/block_u...
syzygy/agent/asan/block_utils.h:50: // Check if a block is corrupt. This check
the block's metadata and its
On 2014/07/25 15:07:45, chrisha wrote:
> Checks* if ...  This checks* the
> 
> Add a comment about this requiring the block redzones to be readable? (This is
> something we know need to start worrying about as we add page protections...)

Done.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
File syzygy/agent/asan/unittest_util.cc (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.cc:134: memset(buffer, 0, sizeof(buffer));
On 2014/07/25 15:07:45, chrisha wrote:
> ::memset, here and elsewhere.

Done.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.cc:196: }
On 2014/07/25 15:07:45, chrisha wrote:
> ubernit: No need for the {} on these single line loops (here and elsewhere).

Done (via a Regex in this project, so replaced in a bunch of different places)

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.cc:230:
common::AlignUp(reinterpret_cast<size_t>(user_ptr)+user_alloc_size,
On 2014/07/25 15:07:45, chrisha wrote:
> Indent +2

Done.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.cc:233: pos < buffer_align_begin +
asan_alloc_size;
On 2014/07/25 15:07:45, chrisha wrote:
> Align with 'const'

Done.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
File syzygy/agent/asan/unittest_util.h (right):

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.h:324: // A unittest fixture that initialize an
ASan heap that can be used in the
On 2014/07/25 15:07:46, chrisha wrote:
> initializes*

Done.

https://codereview.appspot.com/113510043/diff/40001/syzygy/agent/asan/unittes...
syzygy/agent/asan/unittest_util.h:357: // +--- 0 <= size < 2048 bytes---+---2048
bytes---+--2048 bytes--+
On 2014/07/25 15:07:46, chrisha wrote:
> We should grow this to support an alignment of up to 4096, as that's what
we'll
> need to support the about-to-land page protection features.

Done.

[chrisha, 2014-07-25 16:52:26]

Awesome, lgtm with a single nit

https://codereview.appspot.com/113510043/diff/80001/syzygy/agent/asan/block_u...
File syzygy/agent/asan/block_utils.h (right):

https://codereview.appspot.com/113510043/diff/80001/syzygy/agent/asan/block_u...
syzygy/agent/asan/block_utils.h:50: // Checks if a block is corrupt. This check
the block's metadata and its
This checks* the block's

[Sébastien Marchand, 2014-07-25 17:24:18]

Thanks, committing.

https://codereview.appspot.com/113510043/diff/80001/syzygy/agent/asan/block_u...
File syzygy/agent/asan/block_utils.h (right):

https://codereview.appspot.com/113510043/diff/80001/syzygy/agent/asan/block_u...
syzygy/agent/asan/block_utils.h:50: // Checks if a block is corrupt. This check
the block's metadata and its
On 2014/07/25 16:52:25, chrisha wrote:
> This checks* the block's

Oops, sorry I haven't see this one in the previous round of comments...

[Sébastien Marchand, 2014-07-25 18:13:32]

Committed patchset #4 manually as r2195 (presubmit successful).