Closed Bug 1625873 Opened 5 years ago Closed 5 years ago

2.01 - 2.95% Base Content Explicit (linux1804-64-shippable, windows10-64-shippable) regression on push d9edbbdf2679a44ce23044eecb3a8595d9c01930 (Fri March 27 2020)

Categories

(Core :: Memory Allocator, defect)

defect
Not set
normal

Tracking

()

RESOLVED WONTFIX
mozilla76
Tracking Status
firefox76 --- wontfix

People

(Reporter: alexandrui, Unassigned)

References

(Regression)

Details

(Keywords: perf, perf-alert, regression)

We have detected an awsy regression from push:

https://hg.mozilla.org/integration/autoland/pushloghtml?changeset=d9edbbdf2679a44ce23044eecb3a8595d9c01930

As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

3% Base Content Explicit windows10-64-shippable opt 10,136,405.33 -> 10,435,925.33
3% Base Content Explicit windows10-64-shippable opt 10,148,181.33 -> 10,422,613.33
2% Base Content Explicit linux1804-64-shippable opt 13,203,797.33 -> 13,469,354.67

You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=25511

On the page above you can see an alert for each affected platform as well as a link to a graph showing the history of scores for this test. There is also a link to a treeherder page showing the jobs in a pushlog format.

To learn more about the regressing test(s), please see: https://wiki.mozilla.org/AWSY/Tests

Flags: needinfo?(n.nethercote)
Component: Performance → Memory Allocator
Product: Testing → Core
Version: Version 3 → unspecified

This result is a mixture of expected and unexpected.

  • Expected: the change caused an additional 260 KiB (266,240 B) of memory to be allocated in each process, and this matches fairly closely the sizes of the changes seen above. This memory is never touched, however, so it should not cause an increase in physical memory usage (a.k.a. RSS), which is the most important metric. The extra allocation should only consume address space, but PHC is only enabled on 64-bit platforms where address space is in abundant supply, so this is fine.
  • Unexpected: this additional 260 KiB is allocated via mmap/VirtualAlloc, not with malloc, and there is no memory reporter for it, so I wouldn't have expected it to be included in explicit. This is puzzling to me.
Flags: needinfo?(n.nethercote)

Hi Mike, could you set the priority flag of this bug?

Flags: needinfo?(mh+mozilla)
Flags: needinfo?(mh+mozilla) → needinfo?(n.nethercote)

Memory usage was expected to increase with the change. So even though our memory tracking wasn't expected to catch the increase, I will WONTFIX this.

Status: NEW → RESOLVED
Closed: 5 years ago
Flags: needinfo?(n.nethercote)
Resolution: --- → WONTFIX
Has Regression Range: --- → yes
You need to log in before you can comment on or make changes to this bug.