fix: Avoid reallocating fd table during snapshot restore #4107
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
ShadowCurse
reviewed
Sep 13, 2023
Codecov ReportPatch coverage has no change and project coverage change:
Additional details and impacted files@@ Coverage Diff @@
## main #4107 +/- ##
==========================================
- Coverage 82.34% 82.26% -0.09%
==========================================
Files 225 225
Lines 28445 28474 +29
==========================================
Hits 23424 23424
- Misses 5021 5050 +29
Flags with carried forward coverage won't be shown. Click here to find out more.
☔ View full report in Codecov by Sentry. |
roypat
force-pushed
the
prealloc-fdtable
branch
2 times, most recently
from
b61953e to
9063c9d
Compare
bchalios
reviewed
Sep 13, 2023
roypat
force-pushed
the
prealloc-fdtable
branch
from
9063c9d to
51325e2
Compare
bchalios
previously approved these changes
Sep 14, 2023
ShadowCurse
previously approved these changes
Sep 14, 2023
roypat
force-pushed
the
prealloc-fdtable
branch
4 times, most recently
from
abdf7fe to
c0398f7
Compare
bchalios
requested changes
Sep 14, 2023
roypat
force-pushed
the
prealloc-fdtable
branch
3 times, most recently
from
2ae09be to
65cfa5c
Compare
kalyazin
reviewed
Sep 14, 2023
roypat
force-pushed
the
prealloc-fdtable
branch
from
cf87674 to
bbeb1c9
Compare
kalyazin
reviewed
Sep 14, 2023
Linux maintains a per-process file descriptor table. By default, this table has size 64. Whenever space in this table runs out, the file descriptor table gets reallocated, with its size increased to the next power of two. Firecracker creates a lot of eventfds and timerfds for its devices. These are created in the hot path of snapshot restore. For medium to large microVMs, firecracker uses more than 64 file descriptors, meaning we reallocate the file descriptor table on the hot path. However, this reallocation uses a RCU, meaning the kernel needs to hold a lock. To acquire this lock, it needs to wait for all other accesses to the file descriptors to cease, which introduces a severe latency to snapshot-restore times (between 30ms and 70ms). This commit avoids that latency by ensuring the file descriptor table is large enough to hold the jailer-defined limit of file descriptors at process start already. This avoids reallocating the file descriptor table at runtime (and the memory overhead is negligable, as each entry in the fdtable is simply a pointer). Signed-off-by: Patrick Roy <[email protected]>
roypat
force-pushed
the
prealloc-fdtable
branch
from
e19f027 to
e9b07f8
Compare
kalyazin
approved these changes
Sep 14, 2023
ShadowCurse
approved these changes
Sep 14, 2023
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.
Linux maintains a per-process file descriptor table. By default, this table has size 64. Whenever space in this table runs out, the file descriptor table gets reallocated, with its size increased to the next power of two.
Firecracker creates a lot of eventfds and timerfds for its devices. These are created in the hot path of snapshot restore. For medium to large microVMs, firecracker uses more than 64 file descriptors, meaning we reallocate the file descriptor table on the hot path. However, this reallocation uses a RCU, meaning the kernel needs to hold a lock. To acquire this lock, it needs to wait for all other accesses to the file descriptors to cease, which introduces a severe latency to snapshot-restore times (between 30ms and 70ms).
This commit avoids that latency by ensuring the file descriptor table is large enough to hold the jailer-defined limit of file descriptors at process start already. This avoids reallocating the file descriptor table at runtime (and the memory overhead is negligable, as each entry in the fdtable is simply a pointer).
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
CHANGELOG.md.TODOs link to an issue.rust-vmm.