Issue metadata
Sign in to add a comment
|
16 Kib regression in resource_sizes (MonochromePublic.apk) at 469524:469524 |
||||||||||||||||||||
Issue descriptionCaused by: Add support for single sample metrics. https://codereview.chromium.org/2687583002
,
May 8 2017
Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8980171884371570272
,
May 9 2017
,
May 25 2017
,
May 25 2017
,
May 25 2017
dalecurtis@ - sorry for the delay. I noticed that there was already discussion on your CL about the size increase for cronet - your CL also increased binary size on Android by 16 kb. Can you take a quick look at the diff below? Feel free to close this one out if nothing jumps out at you as actionable:
Section Sizes (Total=6.29kb (6444 bytes)):
.bss: 32 bytes (32 bytes) (not included in totals)
.data: 0 bytes (0 bytes) (0.0%)
.data.rel.ro: 0 bytes (0 bytes) (0.0%)
.data.rel.ro.local: 288 bytes (288 bytes) (4.5%)
.rodata: 544 bytes (544 bytes) (8.4%)
.text: 5.48kb (5612 bytes) (87.1%)
89 symbols added (+), 10 changed (~), 5 removed (-), 373967 unchanged (not shown)
5 paths added, 0 removed, 3 changed
Showing 104 symbols (97 unique) with total pss: 6192 bytes
.text=5.23kb .rodata=544 bytes .data*=288 bytes .bss=32 bytes total=6.05kb
Number of unique paths: 8
Index, Running Total, Section@Address, PSS
------------------------------------------------------------
+ 0) 488 (7.9%) t@0x9e1024 488 components/metrics/public/interfaces/single_sample_metrics.mojom.cc
metrics::mojom::SingleSampleMetricsProviderStubDispatch::Accept
+ 1) 872 (14.1%) t@0x2210748 384 content/renderer/render_thread_impl.cc
content::CreateSingleSampleMetricsProvider
+ 2) 1244 (20.1%) t@0x9e00c8 372 components/metrics/single_sample_metrics.cc
mojo::StrongBinding::StrongBinding
+ 3) 1616 (26.1%) t@0x9e02fc 372 components/metrics/single_sample_metrics.cc
mojo::StrongBinding::StrongBinding
+ 4) 1976 (31.9%) t@0x9e0ebc 360 components/metrics/public/interfaces/single_sample_metrics.mojom.cc
metrics::mojom::SingleSampleMetricsProviderProxy::AcquireSingleSampleMetric
~ 5) 2318 (37.4%) r@Group 342 {{no path}}
** merge strings (count=6)
+ 6) 2618 (42.3%) t@0xd3f86c 300 content/browser/renderer_host/render_process_host_impl.cc
service_manager::CallbackBinder::BindInterface
+ 7) 2886 (46.6%) t@0x9e0db0 268 components/metrics/public/interfaces/single_sample_metrics.mojom.cc
metrics::mojom::SingleSampleMetricStubDispatch::Accept
+ 8) 3136 (50.6%) t@0x9e080c 250 components/metrics/single_sample_metrics_factory_impl.cc
metrics::SingleSampleMetricImpl::SetSample
+ 9) 3374 (54.5%) t@0x9e09ec 238 components/metrics/single_sample_metrics_factory_impl.cc
metrics::SingleSampleMetricsFactoryImpl::GetProvider
+ 10) 3606 (58.2%) t@0x9e0904 232 components/metrics/single_sample_metrics_factory_impl.cc
mojo::internal::InterfacePtrState::ConfigureProxyIfNecessary
+ 11) 3812 (61.6%) t@0x9e0adc 206 components/metrics/single_sample_metrics_factory_impl.cc
metrics::SingleSampleMetricsFactoryImpl::CreateMetric
+ 12) 4004 (64.7%) t@0x9e023c 192 components/metrics/single_sample_metrics.cc
metrics::MojoSingleSampleMetricsProvider::AcquireSingleSampleMetric
+ 13) 4172 (67.4%) t@0x9e0470 168 components/metrics/single_sample_metrics.cc
metrics::CreateSingleSampleMetricsProvider
~ 14) 4332 (70.0%) t@0xd4354c 160 content/browser/renderer_host/render_process_host_impl.cc
content::RenderProcessHostImpl::RegisterMojoInterfaces
+ 15) 4488 (72.5%) t@0x9e0c60 156 components/metrics/public/interfaces/single_sample_metrics.mojom.cc
metrics::mojom::SingleSampleMetricRequestValidator::Accept
+ 16) 4644 (75.0%) t@0x9e0bc4 156 components/metrics/public/interfaces/single_sample_metrics.mojom.cc
metrics::mojom::SingleSampleMetricsProviderRequestValidator::Accept
+ 17) 4796 (77.5%) t@0x9e120c 152 components/metrics/public/interfaces/single_sample_metrics.mojom-shared.cc
metrics::mojom::internal::SingleSampleMetricsProvider_AcquireSingleSampleMetric_Params_Data::Validate
~ 18) 4888 (78.9%) t@0x22137fc 92 content/renderer/render_thread_impl.cc
,
May 25 2017
Report seems to indicate this added ~6.29kb, so not the entirety of the 16kb regression shown. ~6kb for a new Mojo service is unfortunately not crazy, see issue 597125 . Seems like WontFix to me?
,
May 25 2017
Actually, I guess there's only one change on that graph for the regression node, so I don't understand the output of the tool in c#9. I don't see anything that stands out as adding unnecessary size though.
,
May 25 2017
Yeah sorry for the confusion - the 16kb regression is for your CL but we use a normalized stat which isn't always super accurate. However, in this case, the tool isn't capturing the entirety of the diff (see issue 726151 ). The 16 kb is almost correct; adding up the section sizes reported by resource_sizes came to just under 16 kb (incl. 8kb of relocations). Regardless, if you don't see anything we can change, I'll close this one out. Thanks for looking into it! |
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by kraynov@chromium.org
, May 8 2017