Closed Bug 1119363 Opened 5 years ago Closed 5 years ago
_bug732665 .xul | Chrome should be invokable from content script with an exhausted stack: 10
04:01:24 INFO - 1516 INFO TEST-START | js/xpconnect/tests/chrome/test_bug732665.xul 04:01:24 INFO - 1517 INFO TEST-PASS | js/xpconnect/tests/chrome/test_bug732665.xul | Chrome should be able to have at least 10 heavy frames more stack than content: 80, 69 04:01:24 INFO - 1518 INFO TEST-UNEXPECTED-FAIL | js/xpconnect/tests/chrome/test_bug732665.xul | Chrome should be invokable from content script with an exhausted stack: 10 - expected PASS 04:01:24 INFO - 1519 INFO MEMORY STAT vsize after test: 1080451072 04:01:24 INFO - 1520 INFO MEMORY STAT vsizeMaxContiguous after test: 8519033622528 04:01:24 INFO - 1521 INFO MEMORY STAT residentFast after test: 360374272 04:01:24 INFO - 1522 INFO MEMORY STAT heapAllocated after test: 107745478 04:01:24 INFO - 1523 INFO TEST-OK | js/xpconnect/tests/chrome/test_bug732665.xul | took 38ms
This means that the stack limits in XPCJSRuntime.cpp need to be adjusted. I recall shu being interested in this code - shu, do you have a few cycles to do something intelligently here?
Not this week, but I'll put it on my queue of things to look at.
Trying 140k instead of the current 120k: https://treeherder.mozilla.org/#/jobs?repo=try&revision=348274581561
Not sure what else is there to be done here. See the try push in comment 31. I forced PGO by adding mk_add_options MOZ_PGO=1. The opt m-oth tests are all green, and the intermittent oranges in the debug seem unrelated to stack exhaustion.
Attachment #8555473 - Flags: review?(bobbyholley) → review+
Status: NEW → RESOLVED
Closed: 5 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla38
Please request Aurora approval on this when you get a chance.
Comment on attachment 8555473 [details] [diff] [review] Tweak trusted stack buffer to give Win64 PGO more room. Approval Request Comment [Feature/regressing bug #]: No clue. [User impact if declined]: Inability to call back into some chrome code when stack is exhausted. [Describe test coverage new/current, TreeHerder]: On central. [Risks and why]: Low. Lowers the amount of content-usable stack space on Win64, but that number is already different on each platform so nobody should depend on it. [String/UUID change made/needed]: None.
Attachment #8555473 - Flags: approval-mozilla-aurora?
Comment on attachment 8555473 [details] [diff] [review] Tweak trusted stack buffer to give Win64 PGO more room. Aurora+
Attachment #8555473 - Flags: approval-mozilla-aurora? → approval-mozilla-aurora+
You need to log in before you can comment on or make changes to this bug.