Message ID | 20190719172927.18181-1-jakub.kicinski@netronome.com |
---|---|
Headers | show |
Series | sockmap/tls fixes | expand |
On Fri, 19 Jul 2019 10:29:13 -0700, Jakub Kicinski wrote: > John says: > > Resolve a series of splats discovered by syzbot and an unhash > TLS issue noted by Eric Dumazet. Sorry for the delay, this code is quite tricky. According to my testing TLS SW and HW should now work, I hope I didn't regress things on the sockmap side. This is not solving all the issues (ugh), apart from HW needing the unhash/shutdown treatment, as discussed we may have a sender stuck in wmem wait while we free context underneath. That's a "minor" UAF for another day..
On 7/19/19 7:37 PM, Jakub Kicinski wrote: > On Fri, 19 Jul 2019 10:29:13 -0700, Jakub Kicinski wrote: >> John says: >> >> Resolve a series of splats discovered by syzbot and an unhash >> TLS issue noted by Eric Dumazet. > > Sorry for the delay, this code is quite tricky. According to my testing > TLS SW and HW should now work, I hope I didn't regress things on the > sockmap side. Applied, thanks everyone!
Jakub Kicinski wrote: > On Fri, 19 Jul 2019 10:29:13 -0700, Jakub Kicinski wrote: > > John says: > > > > Resolve a series of splats discovered by syzbot and an unhash > > TLS issue noted by Eric Dumazet. > > Sorry for the delay, this code is quite tricky. According to my testing > TLS SW and HW should now work, I hope I didn't regress things on the > sockmap side. I'll run it through our CI as well but looks good to me. Thanks a lot for getting this finished up. > > This is not solving all the issues (ugh), apart from HW needing the > unhash/shutdown treatment, as discussed we may have a sender stuck in > wmem wait while we free context underneath. That's a "minor" UAF for > another day.. Agreed. But this should solve most of the syzbot issues and a few crashes we saw in testing real workloads. Thanks, John
Daniel Borkmann wrote: > On 7/19/19 7:37 PM, Jakub Kicinski wrote: > > On Fri, 19 Jul 2019 10:29:13 -0700, Jakub Kicinski wrote: > >> John says: > >> > >> Resolve a series of splats discovered by syzbot and an unhash > >> TLS issue noted by Eric Dumazet. > > > > Sorry for the delay, this code is quite tricky. According to my testing > > TLS SW and HW should now work, I hope I didn't regress things on the > > sockmap side. > > Applied, thanks everyone! Thanks Jakub, for the patches without my signed-off already Acked-by: John Fastabend <john.fastabend@gmail.com>