Message ID | 20211103204736.248403-1-john.fastabend@gmail.com (mailing list archive) |
---|---|
Headers | show |
Series | bpf, sockmap: fixes stress testing and regression | expand |
On Wed, Nov 03, 2021 at 09:47 PM CET, John Fastabend wrote: > Attached are 5 patches that fix issues we found by either stress testing > or updating our CI to LTS kernels. > > Thanks to Jussi for all the hard work tracking down issues and getting > stress testing/CI running. > > First patch was suggested by Jakub to ensure sockets in CLOSE state > were safe from helper side. > > Next two patches are issues discovered by Jussi after writing a stess > testing tool. > > The last two fix an issue noticed while reviewing patches and xlated > code paths also discovered by Jussi. > > v2: Add an initial patch to make sockmap helpers safe with CLOSE > sockets in sockmap > Added Jussi's tested-by line he tested the original patch series. > > John Fastabend (4): > bpf, sockmap: Use stricter sk state checks in sk_lookup_assign > bpf, sockmap: Remove unhash handler for BPF sockmap usage > bpf, sockmap: Fix race in ingress receive verdict with redirect to > self > bpf: sockmap, strparser, and tls are reusing qdisc_skb_cb and > colliding > > Jussi Maki (1): > bpf, sockmap: sk_skb data_end access incorrect when src_reg = dst_reg > > include/linux/skmsg.h | 12 ++++++++ > include/net/strparser.h | 20 +++++++++++- > net/core/filter.c | 64 ++++++++++++++++++++++++++++++++++----- > net/core/sock_map.c | 6 ---- > net/ipv4/tcp_bpf.c | 48 ++++++++++++++++++++++++++++- > net/strparser/strparser.c | 10 +----- > 6 files changed, 135 insertions(+), 25 deletions(-) For the series: Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Hello: This series was applied to bpf/bpf.git (master) by Daniel Borkmann <daniel@iogearbox.net>: On Wed, 3 Nov 2021 13:47:31 -0700 you wrote: > Attached are 5 patches that fix issues we found by either stress testing > or updating our CI to LTS kernels. > > Thanks to Jussi for all the hard work tracking down issues and getting > stress testing/CI running. > > First patch was suggested by Jakub to ensure sockets in CLOSE state > were safe from helper side. > > [...] Here is the summary with links: - [bpf,v2,1/5] bpf, sockmap: Use stricter sk state checks in sk_lookup_assign https://git.kernel.org/bpf/bpf/c/40a34121ac1d - [bpf,v2,2/5] bpf, sockmap: Remove unhash handler for BPF sockmap usage https://git.kernel.org/bpf/bpf/c/b8b8315e39ff - [bpf,v2,3/5] bpf, sockmap: Fix race in ingress receive verdict with redirect to self https://git.kernel.org/bpf/bpf/c/c5d2177a72a1 - [bpf,v2,4/5] bpf: sockmap, strparser, and tls are reusing qdisc_skb_cb and colliding https://git.kernel.org/bpf/bpf/c/e0dc3b93bd7b - [bpf,v2,5/5] bpf, sockmap: sk_skb data_end access incorrect when src_reg = dst_reg https://git.kernel.org/bpf/bpf/c/b2c4618162ec You are awesome, thank you!