Thanks for attending the meeting today. As usual, here are my notes.
=== Discussion ===
- working on fc28 packaging
- pr available
- includes some python3 fixes
- not to be included in next release tag :(
- v0.6.0 tagged
- Some rewriter changes, working with Paul
- 4.17 changed some syscall parameters
- some fixes in bcc, when __syscall prefix is seen
- tracepoint structure bugfix
- introduce option to change license option
- bcc release schedule
- want to release within short timeframe of kernel release
- have tried to keep backwards compatible, but still highly dependent on
- xdp offloads update
- working on examples for using the new api/metadata
- perf numbers to be published after that
- currently experimenting with dynamic hw metadata
- e.g.: vlan, csum complete, hash, flow tag
- can hardware provide offsets to tcp data as a new metadata field?
- some WIP patches for pretty printing maps, currently for array maps
- sockhash fixes from fuzzbot
- bugs in locking and ipv6 ports
- Working on RFC for bounded loop support
- need to expand from finding loop induction vars to all induction vars
- llvm likes to build nested loops, and always jumps with a conditional in
the middle, makes finding loop induction variables hard
- also important to add overflow checks
- possible to modify llvm to make verifying easier?
- annotations? how to rely on them for safety...
- introduce rep counter
- more crazy ideas discussed - pass SSA, dom tree info to verifier
- bcc+btf support?
- not yet, still working on dwarf->btf converter
- pahole has some gaps to be able to convert kernel dwarf
- probably good enough though
=== Attendees ===