1739399 - Linux SM Tsan exception : task resolved with "claim_expired" after 30 minutes

Status: RESOLVED FIXED

(Core :: JavaScript: WebAssembly, defect)

Description

[Cristina Cozmuta (:CrissCozmuta)]

We started noticing this on autoland :
https://treeherder.mozilla.org/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&fromchange=80065f480a1f9e99814a89a8f601bad1ad5488a2&searchStr=linux%2Cx64%2Copt%2Cspidermonkey%2Cbuilds%2Cspidermonkey-sm-tsan-linux64%2Fopt%2Ctsan&tochange=0b465bad573c37dc178bd7b5b0f0dcac1a7b4896&selectedTaskRun=Bq5mUcRCQaucCX66FoPr1Q.5

Comment 1

[Lars T Hansen [:lth]]

Yeah, this is probably going to be a problem. The tests here test very large webassembly heaps - i tried to keep it sane but there's no way that any of the tests can allocate less than 4GB RAM, and each file usually contains several tests. If TSAN allocates shadow storage for this to track read/write activity there's going to be enormous memory pressure.

It may be best to back out the previous patch for now and I can try to figure out how not to run these tests in some configurations. Cristina, can you take care of that?

Comment 2

[Cristina Cozmuta (:CrissCozmuta)]

Hi,
I performed the backout. See bellow the link:

https://hg.mozilla.org/integration/autoland/rev/befddf7c6b413b684f5cda04f4878c71e84a4ddf

Comment 3

[Lars T Hansen [:lth]]

Thanks! I think we'll close this then.

Comment 4

[Intermittent Failures Robot]

3 failures in 3030 pushes (0.001 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• autoland: 3

Platform and build breakdown:

• linux64: 3
  • opt: 3

For more details, see:
https://treeherder.mozilla.org/intermittent-failures/bugdetails?bug=1739399&startday=2021-11-01&endday=2021-11-07&tree=all