Split eBPF load and attach for Go programs #1169
Conversation
grcevski
requested review from
mariomac,
marctc and
rafaelroquetto
as code owners
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #1169 +/- ##
==========================================
+ Coverage 76.83% 81.15% +4.31%
==========================================
Files 136 136
Lines 11451 11497 +46
==========================================
+ Hits 8798 9330 +532
+ Misses 2121 1638 -483
+ Partials 532 529 -3
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
| ta.log.Warn("can't unmount pinned root. Try unmounting and removing it manually", "error", err) | ||
| return | ||
| func UnmountBPFFS(pinPath string, log *slog.Logger) { | ||
| if err := unix.Unmount(pinPath, unix.MNT_FORCE); err != nil { |
There was a problem hiding this comment.
what's the use-case behind MNT_FORCE? I wonder if we should be using MNT_DETACH instead...
grcevski
changed the title
| @@ -9,5 +9,4 @@ import ( | |||
| var ( | |||
| pathRoot = tools.ProjectDir() | |||
| pathOutput = path.Join(pathRoot, "testoutput") | |||
| pathVarRun = path.Join(pathOutput, "run") | |||
There was a problem hiding this comment.
why has this become a problem? Not sure if related, but I noticed a few of my PRs are failing a few tests as well
There was a problem hiding this comment.
Since we'll be removing the BPF file system dependency, I removed all the tracking support for it. Timing is different with the new BPF cleanup and I think we don't always clean it up on time in tests.
This PR changes how we load and attach BPF programs for a few reasons:
Our previous approach was as follows:
The new approach does:
Further work is required to make the same thing happen for the user probe extensions for libssl and nodejs.
Closes #1158