Fix rare TLB refill/invalidate race condition #2534
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
If an SFENCE.VMA with rs1 != x0 or rs2 != x0 happens on the same cycle as an I-TLB refill, the refill still occurs, even if the SFENCE.VMA should've flushed the entry being refilled. SFENCE.VMA with rs1=x0 and rs2=x0 is unaffected. It's exceedingly difficult, but possible, to manifest this bug. We found it by inspection (thanks @khannaudit), not by observing it in the wild.
This signal is sometimes slow to compute, so avoid fanning it out to the clock-enables for the entire TLB. Instead, make only the valid signals depend on its output.
aswaterman
force-pushed
the
fix-sfence-refill-race
branch
from
6111a5f to
898f7c7
Compare
ingallsj
approved these changes
Jun 25, 2020
hcook
pushed a commit
that referenced
this pull request
Aug 4, 2020
Fix rare TLB refill/invalidate race condition
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
If an SFENCE.VMA with rs1 != x0 or rs2 != x0 happens on the same cycle as an I-TLB refill, the refill still occurs, even if the SFENCE.VMA should've flushed the entry being refilled.
SFENCE.VMA with rs1=x0 and rs2=x0 is unaffected.
It's exceedingly difficult to manifest this bug. We found it by inspection (thanks @khannaudit), not by observing it in the wild.