Add Platform.BatteryAbsent histogram to track "missing battery icon" bug

Add Platform.BatteryAbsent histogram to track "missing battery icon" bug

We'll add an upstart job on the Chrome OS side to log whether the failure
is seen on each boot.

BUG=458878
R=jrbarnette@chromium.org,vapier@chromium.org,semenzato@chromium.org,olofj@chromium.org,ejcaruso@chromium.org,isherman@chromium.org

Signed-off-by: Kevin Cernekee <cernekee@chromium.org>;

Committed: https://crrev.com/772156412f85c3aeed361dd84edb7556e56aa3c3
Cr-Commit-Position: refs/heads/master@{#330863}

[Ilya Sherman, 2015-05-20 00:12:18]

LGTM

https://codereview.chromium.org/1133493008/diff/1/tools/metrics/histograms/hi...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/1133493008/diff/1/tools/metrics/histograms/hi...
tools/metrics/histograms/histograms.xml:26901: +<histogram
name="Platform.BatteryAbsent" enum="Boolean">
nit: Could you please define a BooleanAbsent or BooleanPresent enum?  Thanks.

[Kevin Cernekee, 2015-05-20 00:23:11]

https://codereview.chromium.org/1133493008/diff/1/tools/metrics/histograms/hi...
File tools/metrics/histograms/histograms.xml (right):

https://codereview.chromium.org/1133493008/diff/1/tools/metrics/histograms/hi...
tools/metrics/histograms/histograms.xml:26901: +<histogram
name="Platform.BatteryAbsent" enum="Boolean">
On 2015/05/20 00:12:18, Ilya Sherman wrote:
> nit: Could you please define a BooleanAbsent or BooleanPresent enum?  Thanks.

I'll use BooleanAbsent for v2.

The other option is to make it a tristate, representing the three possible log
states:

BAD (at least for laptops): ACPI: Battery Slot [BAT0] (battery absent)
GOOD: ACPI: Battery Slot [BAT0] (battery present)
UNKNOWN: neither message appears

I don't think this is necessary, but others may want to weigh in if they feel
strongly about it.

[semenzato, 2015-05-20 00:59:22]

On 2015/05/20 00:23:11, Kevin Cernekee wrote:
>
https://codereview.chromium.org/1133493008/diff/1/tools/metrics/histograms/hi...
> File tools/metrics/histograms/histograms.xml (right):
> 
>
https://codereview.chromium.org/1133493008/diff/1/tools/metrics/histograms/hi...
> tools/metrics/histograms/histograms.xml:26901: +<histogram
> name="Platform.BatteryAbsent" enum="Boolean">
> On 2015/05/20 00:12:18, Ilya Sherman wrote:
> > nit: Could you please define a BooleanAbsent or BooleanPresent enum? 
Thanks.
> 
> I'll use BooleanAbsent for v2.
> 
> The other option is to make it a tristate, representing the three possible log
> states:
> 
> BAD (at least for laptops): ACPI: Battery Slot [BAT0] (battery absent)
> GOOD: ACPI: Battery Slot [BAT0] (battery present)
> UNKNOWN: neither message appears
> 
> I don't think this is necessary, but others may want to weigh in if they feel
> strongly about it.

I don't know if the tri-state is necessary, i.e. how likely the third state is. 
You can decide.

I would have preferred the value to be BatteryPresent, as "absent" seems like a
negative, and double negatives (not absent) aren't the best choice, but this is
fine.

[Kevin Cernekee, 2015-05-20 01:09:11]

On 2015/05/20 00:59:22, semenzato wrote:
> I would have preferred the value to be BatteryPresent, as "absent" seems like
a
> negative, and double negatives (not absent) aren't the best choice, but this
is
> fine.

Hmm, should we scan for "battery present" in the kernel log instead?  "battery
absent" is being treated as the error indication.  It isn't necessarily the
opposite of "battery present."

If we use a straight boolean, it might be clearer that the stat is answering the
question: "did we find evidence of the BatteryAbsent problem on this boot?"

[Kevin Cernekee, 2015-05-20 22:47:29]

Hi Ilya,

Can we approve/reject this version today?  The sooner we can push an updated
histograms.xml, the sooner I can start testing the client side.

Thanks!

[Ilya Sherman, 2015-05-20 23:08:11]

On 2015/05/20 22:47:29, Kevin Cernekee wrote:
> Hi Ilya,
> 
> Can we approve/reject this version today?  The sooner we can push an updated
> histograms.xml, the sooner I can start testing the client side.
> 
> Thanks!

LGTM.  Sorry, I didn't realize that you were waiting on me.  For future
reference, my LGTMs are meant to be sticky across patch sets, unless I or the CL
author specifically ask for another round of review.

[Kevin Cernekee, 2015-05-20 23:11:02]

The CQ bit was checked by cernekee@chromium.org

[commit-bot: I haz the power, 2015-05-20 23:11:41]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1133493008/20001

[commit-bot: I haz the power, 2015-05-21 00:43:32]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-05-21 00:44:18]

Patchset 2 (id:??) landed as
https://crrev.com/772156412f85c3aeed361dd84edb7556e56aa3c3
Cr-Commit-Position: refs/heads/master@{#330863}