minutes: IO Visor TSC/Dev Meeting
Thanks for dialing in to the call today. Here are notes from the discussion.
=== Discussion ===
* Spinlock patches being worked on
* Proposal to standardize the BPF ELF format/api to that used by libbpf/bpftool
- First user might be iproute2 conversion
- Should support multi-language bindings (go)
* Tagged bcc release
* bringing libbpf into bcc
- few related bugs
* bringing BTF support into bcc - WIP
- issue on getting remapped file (from compilation stage) info into bcc
- getting btf type during map creation in right order in bcc...needs some
* build id support added - first user in trace.py
* respin jmp32 patch set - v2
- implementation in working state
- isa for subregister looking good
- codegen giving smaller results (num of instructions)
- will resend patch set after fixing test failure
* patches to be sent out for sockmap and ktls
* the parser in the filter on the ktls rx side needs some work
- there are some edge cases between ktls and other parsers
* refactoring test cases for verifier improvements
* Helping to refactor/review test verifier code
- splitting up tests by category
* Cilium work to add in second datapath (ipvlan based)
=== Attendees ===
On 24/01/2019 02:33, Brenden Blanco wrote:
Jiong:Finished re-spin and rebased on top of latest bpf-next.
My set will potentially conflict with the following three other bpf-next
pending patches/sets on selftests Makefile and test_verifier.c:
- Stanislav's LLVM_READELF change on Makefile.
- Alexei's spin_lock change on test_verifier.c.
- Jakub's test_verifier.c refactor on test_verifier.c.
The conflict should be easy to resolve. For test_verifier.c, my set just
adds new tests at the end of the test array. I will send out v3 shortly.