Messages in this thread | | | Date | Thu, 13 Jan 2022 16:23:59 +0800 | Subject | Re: [PATCH net] net/smc: Avoid setting clcsock options after clcsock released | From | Wen Gu <> |
| |
Thanks for your reply.
On 2022/1/12 5:38 pm, Karsten Graul wrote: > On 11/01/2022 17:34, Wen Gu wrote: >> Thanks for your reply. >> >> On 2022/1/11 6:03 pm, Karsten Graul wrote: >>> On 10/01/2022 10:38, Wen Gu wrote: >>>> We encountered a crash in smc_setsockopt() and it is caused by >>>> accessing smc->clcsock after clcsock was released. > > I like the idea to use RCU with rcu_assign_pointer() to protect that pointer! > > Lets go with your initial patch (improved to address the access in smc_switch_to_fallback()) > for now because it solves your current problem. >
OK, I will improve the patch, adding check before clcsock access in smc_switch_to_fallback() and return an error (-EBADF) if smc->clcsock is NULL. The caller of smc_switch_to_fallback() will check the return value to identify whether fallback is possible.
> I put that RCU thing on our list, but if either of us here starts working on that please let the > others know so we don't end up doing parallel work on this. But I doubt that we will be able to start working > on that soon. > > Thanks for the good idea!
Thank you! If I start working on the RCU things, I will send a RFC to let you know.
Thanks, Wen Gu
| |