No; I ran reftests in a local TSAN build under `rr` and wasn't yet able to reproduce. So at this point, this still seems likely to be a bug in TSAN itself that we don't have any special insight into, though we might learn more if anyone is able to catch it in a debugger (as I've tried to do, so far unsuccessfully). Worst-case, if this is causing sheriff time to be wasted classifying failures, I wonder if we could consider this to be a blue "infrastructure issue" type bug (considering TSAN to be part of the infrastructure on these test runs), and write some sort of log-processing rule to detect this and automatically categorize it as such? (And automatically retrigger the task, as we try to do for other infra failures.)
Bug 1770595 Comment 76 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
No; I ran reftests in a local TSAN build under `rr` for about a day and wasn't yet able to reproduce. So at this point, this still seems likely to be a bug in TSAN itself that we don't have any special insight into, though we might learn more if anyone is able to catch it in a debugger (as I've tried to do, so far unsuccessfully). Worst-case, if this is causing sheriff time to be wasted classifying failures, I wonder if we could consider this to be a blue "infrastructure issue" type bug (considering TSAN to be part of the infrastructure on these test runs), and write some sort of log-processing rule to detect this and automatically categorize it as such? (And automatically retrigger the task, as we try to do for other infra failures.)
No; I ran reftests in a local TSAN build under `rr` for about a day and wasn't yet able to reproduce. So at this point, this still seems likely to be a bug (or weird edge case) in TSAN itself that we don't have any special insight into, though we might learn more if anyone is able to catch it in a debugger (as I've tried to do, so far unsuccessfully). Worst-case, if this is causing sheriff time to be wasted classifying failures, I wonder if we could consider this to be a blue "infrastructure issue" type bug (considering TSAN to be part of the infrastructure on these test runs), and write some sort of log-processing rule to detect this and automatically categorize it as such? (And automatically retrigger the task, as we try to do for other infra failures.)