-
Notifications
You must be signed in to change notification settings - Fork 715
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Enable paranoid opt. check for jitdumps resulting from crashes in the optimizer #9386
Comments
This issue has been encountered in #8992 where developers requested that paranoid opt checks be enabled for a crash in area of usedef analysis. |
@andrewcraik @cathyzhyi @vijaysun-omr any thoughts on this? |
It seems a reasonable thing to do from my perspective. |
Sounds reasonable to me as well. |
Yes, seems like a valid place for paranoidOptCheck to run...it does run the risk of "getting in the way" of the original crash but surely we would'nt have two independent bugs in the same compile and so using the option to catch it as early as possible seems right. |
I'd like to propose that we enable the
TR_EnableParanoidOptCheck
option when we are generating a recompilation jitdump resulting from a crash whose VMState determines we are somewhere in the optimizer.Personally I think this change should be made as I don't see too much of a downside to it, and a large benefit that can potentially identify structural problems found in the trees/CFG before an optimization runs. This can further aid the developers by having an assumption that the tree shapes, reference counts, and the CFG are valid before an optimization starts.
Before making this change I'd like to get input from resident experts in the area of the optimizer for their thoughts, so subscribing @andrewcraik @cathyzhyi @vijaysun-omr for input.
The text was updated successfully, but these errors were encountered: