Cherry-pick #15771 to 7.x: Socket dataset: Workaround for bogus dereference in kernel 5.x #15791
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
…ic#15771) This is a tentative workaround for the problems in Auditbeat's system/socket dataset when run under 5.x kernels. On older kernels, we could rely on dereferencing a NULL or invalid pointer returning zeroed memory. However, seems that in the tested 5.x kernels is not the case. Dereferencing a NULL pointer returns bogus memory, which causes some wrong codepaths to be taken in a couple of kprobes defined by the dataset. This so far seems only to affect udp_sendmsg and udpv6_sendmsg, which caused it to attribute traffic to bogus IP addresses. In turn this caused the test-connected-udp system tests to fail. (cherry picked from commit 0dab517)
adriansr
force-pushed
the
backport_15771_7.x
branch
from
5339587
to
d1145ea
Compare
andrewkroh
approved these changes
Jan 25, 2020
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Outdated suggestions cannot be applied.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.
Cherry-pick of PR #15771 to 7.x branch. Original message:
This is a tentative workaround for the problems in Auditbeat's system/socket dataset when run under 5.x kernels.
On older kernels, we could rely on dereferencing a NULL or invalid pointer returning zeroed memory. However, seems that in the tested 5.x kernels is not the case. Dereferencing a NULL pointer returns bogus memory, which causes some wrong codepaths to be taken in a couple of
kprobes defined by the dataset.
This so far seems only to affect udp_sendmsg and udpv6_sendmsg, which caused it to attribute traffic to bogus IP addresses. In turn this caused the test-connected-udp system tests to fail.
In parallel I'm trying to find the reason for this different behavior and ideally have the tracing subsystem behave like in previous versions, so this patch is not needed.