nvproxy: keep uvmFDMemmapFile.pfm refcount up to date#12657
Merged
copybara-service[bot] merged 1 commit intomasterfrom Feb 27, 2026
Merged
nvproxy: keep uvmFDMemmapFile.pfm refcount up to date#12657copybara-service[bot] merged 1 commit intomasterfrom
copybara-service[bot] merged 1 commit intomasterfrom
Conversation
copybara-service
bot
force-pushed
the
test/cl875885672
branch
from
ea1805e to
a7f85aa
Compare
copybara-service
bot
changed the title
copybara-service
bot
force-pushed
the
test/cl875885672
branch
6 times, most recently
from
a8d969a to
d426c0c
Compare
Otherwise, mappings owned by the PreciseHostFileMapper leak after the uvmFD is released, interfering with future mappings of the same nvidia-uvm addresses. Without platform/kvm, hitting this requires the application workload to touch the same `cudaMallocManaged` memory addresses with syscalls (e.g. read/write) using two distinct uvmFDs (e.g. in two separate sequentially-executed process trees). On platform/kvm, hitting this only requires the application to *allocate* the same `cudaMallocManaged` memory addresses using two distinct uvmFDs, but use of nvidia-uvm on platform/kvm is discouraged due to #11436. PiperOrigin-RevId: 876436522
copybara-service
bot
force-pushed
the
test/cl875885672
branch
from
d426c0c to
a12250c
Compare
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
nvproxy: keep uvmFDMemmapFile.pfm refcount up to date
Otherwise, mappings owned by the PreciseHostFileMapper leak after the uvmFD is
released, interfering with future mappings of the same nvidia-uvm addresses.
Without platform/kvm, hitting this requires the application workload to touch
the same
cudaMallocManagedmemory addresses with syscalls (e.g. read/write)using two distinct uvmFDs (e.g. in two separate sequentially-executed process
trees). On platform/kvm, hitting this only requires the application to
allocate the same
cudaMallocManagedmemory addresses using two distinctuvmFDs, but use of nvidia-uvm on platform/kvm is discouraged due to #11436.