Add basic support for CAPS exe

Add basic support for CAPS exe

1- A singleton
2- A logger

Now CAPS depends on base and on chrome/common_version. About 190KB
on plain release build.

BUG=447073
Committed: https://crrev.com/51747bc6af43f662054ebee3a73976edad07f4a5
Cr-Commit-Position: refs/heads/master@{#314411}

[cpu_(ooo_6.6-7.5), 2015-02-01 19:08:40]

cpu@chromium.org changed reviewers:
+ scottmg@chromium.org

[cpu_(ooo_6.6-7.5), 2015-02-01 19:08:41]

ptal

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/caps.gyp (left):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/caps.gyp:42: '../..',
I think I cargo cult this.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/caps.gyp (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/caps.gyp:56:
'<(DEPTH)/chrome/chrome.gyp:common_version',
that is how I force chrome/common/chrome_version_info_values.h

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/logger_win.cc (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:16: 
my loggers are never free form, that is, you can always write a log reader
without needing AI, thus the table.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:55: }
the [24] = 0 is because stupid thing puts \n\0 always.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:64: // for anything else.
I am having bruce change our logging to use This One Weird Trick to make Your
Logging Awesome.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/logger_win.h (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.h:15: 
sorry for the reinvention of the wheel. The one in base is a square wheel made
of glass.

[scottmg, 2015-02-01 19:55:36]

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/caps.gyp (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/caps.gyp:43: '<(DEPTH)/',
don't use DEPTH in gyp files. it's only meant for gypi files (where they can be
relative to different places so you can't say "../../" or whatever.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/caps.gyp:55: '<(DEPTH)/base/base.gyp:base',
same here, just ../../../base/base.gyp:base is preferred.

crashpad's would probably be better eventually as it drops most of the goop and
just keeps a minimal useful set. also, it doesn't change unless we change it.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/caps.gyp:56:
'<(DEPTH)/chrome/chrome.gyp:common_version',
On 2015/02/01 19:08:41, cpu wrote:
> that is how I force chrome/common/chrome_version_info_values.h

Acknowledged.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/logger_win.cc (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:15: const char kMessageHeader[] =
"%03d~ %04x %s ~\n";
GetTickCount will normally be larger than 4 digits, either remove the 04, or
make it larger?

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:16: 
On 2015/02/01 19:08:41, cpu wrote:
> my loggers are never free form, that is, you can always write a log reader
> without needing AI, thus the table.

until version has a ')' in it!

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:34: bool
WriteFormatedLogLine(HANDLE file, size_t index, ...) {
Formated -> Formatted, and in comments

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:52: asctime_s(time_buf,
arraysize(time_buf), &local_time);
you can just remove the arraysize arg, as there's a [N] overload in C++.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:63: // opening a file like so
allows atomic appends, but can't be used
capital Opening

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:64: // for anything else.
On 2015/02/01 19:08:41, cpu wrote:
> I am having bruce change our logging to use This One Weird Trick to make Your
> Logging Awesome.

Nice, hadn't used that before.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:65: auto kShareAll =
FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE;
I find all these auto's a little frightening, but I guess in this case if it
compiles it's OK.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:70: NULL);
NULL->nullptr

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:79: WriteFormatedLogLine(file_, 1,
::GetCurrentProcessId(), DateTime().c_str());
The 0, 1, etc. seem a bit non-obvious. I don't understand who will be calling
this, but I guess something else needs to go in the header later?

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/main_win.cc (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/main_win.cc:21: // what directory we write
depends if we are being run from the actual
capital at start of comments

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/main_win.cc:26: // start logging.
Do you want the logging start and the singleton check in this order? Will it be
able to open the same log for append in this order?

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/process_singleton_win.cc (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/process_singleton_win.cc:11:
ProcessSingleton::ProcessSingleton() : mutex_(NULL) {
NULL -> nullptr

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/process_singleton_win.cc:12: auto mutex =
::CreateMutexW(nullptr, TRUE, L"CHROME.CAPS.V1");
no W (are we not building as _UNICODE properly? if so, we should fix that)

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/return_codes.h (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/return_codes.h:9: // Exit codes for CAPS. Don't
reorder these values.
add new line above (or remove the one at the bottom)

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/return_codes.h:10: enum ReturnCodes {
ExitCodes (per your comment) seems like a better name (and RC->EC). I'd probably
go with EXIT_CODE_NORMAL_EXIT, ... as RC_ made me thing of "resource compiler"
initially.

[cpu_(ooo_6.6-7.5), 2015-02-03 02:44:59]

ptal

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/caps.gyp (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/caps.gyp:43: '<(DEPTH)/',
On 2015/02/01 19:55:35, scottmg wrote:
> don't use DEPTH in gyp files. it's only meant for gypi files (where they can
be
> relative to different places so you can't say "../../" or whatever.

Done.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/caps.gyp:55: '<(DEPTH)/base/base.gyp:base',
On 2015/02/01 19:55:35, scottmg wrote:
> same here, just ../../../base/base.gyp:base is preferred.
> 
> crashpad's would probably be better eventually as it drops most of the goop
and
> just keeps a minimal useful set. also, it doesn't change unless we change it.

Done.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/logger_win.cc (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:15: const char kMessageHeader[] =
"%03d~ %04x %s ~\n";
On 2015/02/01 19:55:35, scottmg wrote:
> GetTickCount will normally be larger than 4 digits, either remove the 04, or
> make it larger?

Done.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:16: 
On 2015/02/01 19:55:35, scottmg wrote:
> On 2015/02/01 19:08:41, cpu wrote:
> > my loggers are never free form, that is, you can always write a log reader
> > without needing AI, thus the table.
> 
> until version has a ')' in it!

Acknowledged.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:34: bool
WriteFormatedLogLine(HANDLE file, size_t index, ...) {
On 2015/02/01 19:55:35, scottmg wrote:
> Formated -> Formatted, and in comments

Done.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:52: asctime_s(time_buf,
arraysize(time_buf), &local_time);
On 2015/02/01 19:55:35, scottmg wrote:
> you can just remove the arraysize arg, as there's a [N] overload in C++.

Done.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:63: // opening a file like so
allows atomic appends, but can't be used
On 2015/02/01 19:55:35, scottmg wrote:
> capital Opening

Done.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/logger_win.cc:79: WriteFormatedLogLine(file_, 1,
::GetCurrentProcessId(), DateTime().c_str());
On 2015/02/01 19:55:35, scottmg wrote:
> The 0, 1, etc. seem a bit non-obvious. I don't understand who will be calling
> this, but I guess something else needs to go in the header later?

So the idea is that the logger will have methods with GoodNames(tm) so the
numbers only need to be known within this file. We could also have an enum in
this file but seemed more effort.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/main_win.cc (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/main_win.cc:21: // what directory we write
depends if we are being run from the actual
On 2015/02/01 19:55:35, scottmg wrote:
> capital at start of comments

Done.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/main_win.cc:26: // start logging.
On 2015/02/01 19:55:35, scottmg wrote:
> Do you want the logging start and the singleton check in this order? Will it
be
> able to open the same log for append in this order?

Ya, it should be able to handle different processes with atomic writes, I will
test manually and at some point really test.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/process_singleton_win.cc (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/process_singleton_win.cc:12: auto mutex =
::CreateMutexW(nullptr, TRUE, L"CHROME.CAPS.V1");
On 2015/02/01 19:55:35, scottmg wrote:
> no W (are we not building as _UNICODE properly? if so, we should fix that)

afaik we build ok, old habits die hard. changed.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
File chrome/tools/crash_service/caps/return_codes.h (right):

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/return_codes.h:9: // Exit codes for CAPS. Don't
reorder these values.
On 2015/02/01 19:55:35, scottmg wrote:
> add new line above (or remove the one at the bottom)

Done.

https://codereview.chromium.org/890213002/diff/1/chrome/tools/crash_service/c...
chrome/tools/crash_service/caps/return_codes.h:10: enum ReturnCodes {
On 2015/02/01 19:55:35, scottmg wrote:
> ExitCodes (per your comment) seems like a better name (and RC->EC). I'd
probably
> go with EXIT_CODE_NORMAL_EXIT, ... as RC_ made me thing of "resource compiler"
> initially.

Done.

[scottmg, 2015-02-03 04:00:11]

LGTM

[scottmg, 2015-02-03 04:00:53]

https://codereview.chromium.org/890213002/diff/20001/chrome/tools/crash_servi...
File chrome/tools/crash_service/caps/caps.gyp (right):

https://codereview.chromium.org/890213002/diff/20001/chrome/tools/crash_servi...
chrome/tools/crash_service/caps/caps.gyp:55: '<../../../../base/base.gyp:base',
oops, no < here.

[cpu_(ooo_6.6-7.5), 2015-02-03 19:17:22]

The CQ bit was checked by cpu@chromium.org

[commit-bot: I haz the power, 2015-02-03 19:18:07]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/890213002/40001

[commit-bot: I haz the power, 2015-02-03 19:39:46]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-02-03 19:39:47]

Try jobs failed on following builders:
  win_chromium_x64_rel_ng on tryserver.chromium.win
(http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_x64_...)

[cpu_(ooo_6.6-7.5), 2015-02-03 20:19:18]

The CQ bit was checked by cpu@chromium.org

[commit-bot: I haz the power, 2015-02-03 20:20:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/890213002/60001

[commit-bot: I haz the power, 2015-02-03 20:38:51]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-02-03 20:38:52]

Try jobs failed on following builders:
  win_chromium_rel_ng on tryserver.chromium.win
(http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_...)

[scottmg, 2015-02-03 20:43:26]

https://codereview.chromium.org/890213002/diff/60001/chrome/tools/crash_servi...
File chrome/tools/crash_service/caps/logger_win.cc (right):

https://codereview.chromium.org/890213002/diff/60001/chrome/tools/crash_servi...
chrome/tools/crash_service/caps/logger_win.cc:30:
reinterpret_cast<DWORD>(txt.size()),
looks like a real failure, wants static_cast here, not reinterpret_cast (or
base::checked_cast if you're feeling fancy)

[cpu_(ooo_6.6-7.5), 2015-02-03 21:15:48]

The CQ bit was checked by cpu@chromium.org

[commit-bot: I haz the power, 2015-02-03 21:16:44]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/890213002/80001

[commit-bot: I haz the power, 2015-02-03 21:55:33]

Committed patchset #5 (id:80001)

[commit-bot: I haz the power, 2015-02-03 21:56:18]

Patchset 5 (id:??) landed as
https://crrev.com/51747bc6af43f662054ebee3a73976edad07f4a5
Cr-Commit-Position: refs/heads/master@{#314411}