The code for serializing an EventKey in SerializeEventsArray does
something peculiar: it creates a temporary array of nsCStrings, copies
the appropriate strings from the event info for the key, and then
converts those into JavaScript strings. But we shouldn't need to do any
copying into the temporary array. We can do everything directly on the
strings from the event info themselves.
It seemed best to introduce a ToJSString for this purpose, which comes
in handy in a couple other places as well.
This patch adds support for value arrays to the internal_JSHistogram_Add method.
The histogram.add() js method can now be called with a single value or an array of values.
The accumulation is "all-or-nothing", i.e. if there is even a single invalid value in the
array, then nothing is accumulated.
The logic for type checking arguments before accumulation has been moved into a separate
function: internal_JSHistogram_CoerceValue. This function can also be used for
internal_JSKeyedHistogram_* methods in the future. This patch includes logic for clamping
values greater than UINT32_MAX to UINT32_MAX.
Added xpcshell tests in telemetry/tests/unit/test_TelemetryHistograms.js to check functionality.
We need to clamp accumulations to fit in our data representation (int). This
patch records the number of times, and for which probes, we had to do so.
MozReview-Commit-ID: GSs3oFvLKlL
Internally Histograms use int, which is signed.
Our API uses uint32_t, which is not.
Our JS API uses ToUint32, which takes provided values modulo UINT32_MAX.
This patch consolidates behaviour so that all values greater than INT_MAX will
be correctly recorded in the highest bucket of a Histogram... instead of
having values between INT_MAX and UINT32_MAX end up as 0s, and values above
UINT32_MAX being taken modulo UINT32_MAX, but only from the JS API.
MozReview-Commit-ID: CzaaNZkPEji
Telemetry::Accumulate now supports two new signatures: Accumulate(HistogramId, nsTArray stringLabels) and
Accumulate(nsTArray enumValues). In the stringLabels case, if the array contains an invalid label, then no
samples are accumulated at all. In the enumValues array case, the class template ensures that we do not have
a mismatch of labels in the array, since the enumValues class is tied to the id of the histogram we want to
accumulate to.
This switches from `nsDependentCString` to `nsLiteralCString` when adding
histograms to the table. It will help save some memory and should avoid the OOM
smalls we have been seeing.
Extended the Telemetry::Accumulate API to take a Histogram ID, string key, and a nsTArray<uint32_t> of samples.
Test cases check for linear, count and histograms with a set of allowed keys. Made changes to fix try failures.
Added another Telemetry::Accumulate function that takes a histogram id and an array of samples as arguments.
As of this patch, adding multiple samples to keyed and categorical histograms is not supported.
Since LinearHistogram and its descendants inherit ranges_ from
Histogram, and we wanted to replace the copying into a std::vec
for Histogram, the simplest approach seemed to just be to
precompute ranges for all histograms, exponential or otherwise.
This should have the added benefit of reducing the memory
footprint for those histograms, since they will benefit from the
deduplication work that the precomputing script already does.
MozReview-Commit-ID: JTV5Dej5ZIb
Switch from static multi-dimensional arrays to dynamic one-dimensional arrays
that are only allocated in the parent process.
MozReview-Commit-ID: tyGEFhU2Fq
Switch from static multi-dimensional arrays to dynamic one-dimensional arrays
that are only allocated in the parent process.
MozReview-Commit-ID: tyGEFhU2Fq
With the removal of the old addonHistograms, all histograms are now registered.
So removing registered(Keyed)Histograms should be straightforward?
Unfortunately not, as this was how we filtered data based on dataset
(opt-in/opt-out), so a little more fiddling was needed to get C++ to only
serialize dataset-appropriate data (instead of post-facto filtering it in JS).
MozReview-Commit-ID: HDplhmzmzJl
The log and exp calls in base::Histogram::InitializeBucketRange()
were showing up in profiles. This patch uses the precomputed
buckets for exponential histograms instead of computing them at
runtime. Though linear histograms do show up in the profile that
prompted this change, they contribute much less, and due to the
trivial nature of generating these, it's unlikely that a static
cache would provide much if any speedup.
MozReview-Commit-ID: IavFwoWjFhk
The log and exp calls in base::Histogram::InitializeBucketRange()
were showing up in profiles. This patch uses the precomputed
buckets for exponential histograms instead of computing them at
runtime. Though linear histograms do show up in the profile that
prompted this change, they contribute much less, and due to the
trivial nature of generating these, it's unlikely that a static
cache would provide much if any speedup.
MozReview-Commit-ID: IavFwoWjFhk
Previously we assumed count histograms were always present in payloads.
This was an erroneous assumption as count histograms were only 0 if they were
session histograms, or if they were from subsession histograms from subsessions
_after_ a subsession when they held a non-0 value.
So let's just treat count histograms as normal histograms from now on, without
any of this "sometimes 0" nonsense. This simplifies the code, tests, and our
understanding... and _should_ have few/zero downstream effects since the
existing behaviour was so poorly-understood (though exactly tested).
MozReview-Commit-ID: BH108ksygGw
TelemetrySession's getKeyedHistograms asks for each keyed histogram
individually. This is inefficient and doesn't work well with the storage
refactor.
So, plumb through a subsession keyed histogram snapshot API and convert
TelemetrySession over to using it.
MozReview-Commit-ID: Af9dTqw99UA
Previously we were doing bad string manipulation nonsense. Now when asked for
snapshots C++ can return a properly-formated Object tree.
MozReview-Commit-ID: HAvIbgzUvMU
Previously we used the base::StatisticsRecorder object for storage by name.
This is keyed by histogram name, which doesn't match our storage reality anymore.
Instead we use a name to refer to a set of histogram instances that record data from different processes, as well as separating session and subsession data.
In this re-write, we instead introduce the following lookup paths (managed in TelemetryHistogram.cpp):
- Main storage:
- (histogramId, processId, sessionOrSubsession) -> Histogram*
- (histogramId, processId) -> KeyedHistogram* (this handles subsessions internally)
- Lookup:
- (histogramName) -> histogramId
- (HistogramID) -> bool (is recording enabled for this histogram?)
This is wrapped with a few lookup functions.
This also allows us to keep HistogramIDs in the JS histogram instances now, instead of pointers to Histogram instances.
That means Histogram instance life-time management is now properly contained inside TelemetryHistogram.cpp.
MozReview-Commit-ID: 5yijGv7mc89
The Chromium IPC histogram code used the StatisticsRecorder object for storage.
This is keyed by histogram name, which doesn't match our storage reality anymore.
Instead we use a name to refer to a set of histogram instances that record data from different processes, as well as separating session and subsession data.
Consequently we need to rewrite this storage, which means StatisticsRecorder is not used anymore.
MozReview-Commit-ID: 1LC7YubpKaD
With bug 1374419 solved, static asserts in TelemetryHistogram.cpp can be improved.
Specifically, JS::gcreason::NUM_TELEMETRY_REASONS can now be compared against GC_MINOR_REASON, GC_MINOR_REASON_LONG, and GC_REASON_2 in the now constexpr gHistograms. Likewise, mozilla::StartupTimeline::MAX_EVENT_ID can be compared against STARTUP_MEASUREMENT_ERRORS.
This commit modifies the static asserts to perform these comparisons in place of ones against hardcoded values and modifies the accompanying comment to remove the TODO section.
Unfortunately it means using FLAG histograms because I need to test that their
"we have a hidden default value" behaviour is appropriately silenced as well.
MozReview-Commit-ID: J8xGXwCTRIg
Updates the core Telemetry code to use the new headers and support the extension process.
TelemetryHistogram is not cleanly refactored here; doing this uncovered a few time consuming issues with the various lookup code paths.
Updates the core Telemetry code to use the new headers and support the extension process.
TelemetryHistogram is not cleanly refactored here; doing this uncovered a few time consuming issues with the various lookup code paths.
Updates the core Telemetry code to use the new headers and support the extension process.
TelemetryHistogram is not cleanly refactored here; doing this uncovered a few time consuming issues with the various lookup code paths.
addonHistograms isn't being used and has started getting in the way.
So it goes.
Leave the "Addon Histograms" section in about:telemetry for 60 days.
Remove it in bug 1355882
MozReview-Commit-ID: 4lm7ONirofl
addonHistograms isn't being used and has started getting in the way.
So it goes.
Leave the "Addon Histograms" section in about:telemetry for 60 days.
Remove it in bug 1355882
MozReview-Commit-ID: 4lm7ONirofl
This also fixes an existing bug where we checked for the wrong type in JSKeyedHistogram_Add().
KeyedHistogram::GetHistogramType() returns one of the histogram types from nsITelemetry.idl, not one of the chromium types (base::*).
I extended the test coverage accordingly, as it seemed to be lacking.
