fix(aarch64): override fabricated CLIDR_EL1 to match host cache topology#5780
Merged
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #5780 +/- ##
=======================================
Coverage 83.02% 83.02%
=======================================
Files 276 276
Lines 29340 29418 +78
=======================================
+ Hits 24359 24425 +66
- Misses 4981 4993 +12
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
kalyazin
force-pushed
the
arm_topo
branch
2 times, most recently
from
3f6742b to
4184864
Compare
kalyazin
added
the
Status: Awaiting review
JamesC1305
reviewed
Mar 23, 2026
JamesC1305
reviewed
Mar 23, 2026
kalyazin
force-pushed
the
arm_topo
branch
2 times, most recently
from
af73c49 to
1f2e62b
Compare
ShadowCurse
reviewed
Mar 23, 2026
ShadowCurse
previously approved these changes
Mar 23, 2026
JamesC1305
previously approved these changes
Mar 23, 2026
Manciukic
reviewed
Mar 24, 2026
Manciukic
left a comment
Manciukic
left a comment
Contributor
There was a problem hiding this comment.
LGTM, just one nit that we could print a warning in case we fail to read the host cache topology.
Since host kernel 6.3 (commit 7af0c2534f4c), KVM fabricates CLIDR_EL1 instead of passing through the host's real value. On hosts with IDC=1 and DIC=0 (e.g. Neoverse V1), the fabricated CLIDR exposes only L1=Unified when the host actually has separate L1d+L1i, L2, and L3. Guest kernels >= 6.1.156 backported init_of_cache_level() which counts cache leaves from the DT, while populate_cache_leaves() uses CLIDR_EL1. When the DT (built from host sysfs) describes more cache entries than CLIDR_EL1, the mismatch causes cache sysfs entries to not be created, breaking /sys/devices/system/cpu/cpu*/cache/* in the guest. Fix this by reading the current CLIDR_EL1 from vCPU 0, merging in the ctype and LoC fields derived from the host's sysfs cache topology, and writing the result back to each vCPU via KVM_SET_ONE_REG. Fields that cannot be derived from sysfs (LoUU, LoUIS, ICB, Ttype) are preserved from the original CLIDR_EL1. This makes CLIDR_EL1 consistent with the FDT, which already describes the real host caches. On pre-6.3 kernels, KVM passes through the real host CLIDR rather than fabricating one. Since the sysfs cache topology already matches the real CLIDR, the merge produces the same value, the write is skipped, and the override is effectively a no-op. This approach preserves the full host cache information for the guest rather than stripping the FDT to match the fabricated CLIDR. Signed-off-by: Nikita Kalyazin <kalyazin@amazon.com>
ShadowCurse
approved these changes
Mar 24, 2026
JamesC1305
approved these changes
Mar 24, 2026
Manciukic
approved these changes
Mar 24, 2026
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
Since host kernel 6.3 (commit 7af0c2534f4c), KVM fabricates CLIDR_EL1 instead of passing through the host's real value. On hosts with IDC=1 and DIC=0 (e.g. Neoverse V1), the fabricated CLIDR exposes only L1=Unified when the host actually has separate L1d+L1i, L2, and L3.
Guest kernels >= 6.1.156 backported init_of_cache_level() which counts cache leaves from the DT, while populate_cache_leaves() uses CLIDR_EL1. When the DT (built from host sysfs) describes more cache entries than CLIDR_EL1, the mismatch causes cache sysfs entries to not be created, breaking /sys/devices/system/cpu/cpu*/cache/* in the guest.
Fix this by reading the current CLIDR_EL1 from vCPU 0, merging in the ctype and LoC fields derived from the host's sysfs cache topology, and writing the result back to each vCPU via KVM_SET_ONE_REG. Fields that cannot be derived from sysfs (LoUU, LoUIS, ICB, Ttype) are preserved from the original CLIDR_EL1. This makes CLIDR_EL1 consistent with the FDT, which already describes the real host caches.
On pre-6.3 kernels, KVM passes through the real host CLIDR rather than fabricating one. Since the sysfs cache topology already matches the real CLIDR, the merge produces the same value, the write is skipped, and the override is effectively a no-op.
This approach preserves the full host cache information for the guest rather than stripping the FDT to match the fabricated CLIDR.
License Acceptance
By submitting this pull request, I confirm that my contribution is made under
the terms of the Apache 2.0 license. For more information on following Developer
Certificate of Origin and signing off your commits, please check
CONTRIBUTING.md.PR Checklist
tools/devtool checkbuild --allto verify that the PR passesbuild checks on all supported architectures.
tools/devtool checkstyleto verify that the PR passes theautomated style checks.
how they are solving the problem in a clear and encompassing way.
in the PR.
CHANGELOG.md.Runbook for Firecracker API changes.
integration tests.
TODO.rust-vmm.