Re: kprobes, livepatch and FTRACE_OPS_FL_IPMODIFY

2019-07-24 Thread Masami Hiramatsu
Hi Joe, On Thu, 25 Jul 2019 09:32:08 +0900 Masami Hiramatsu wrote: > NO, that flag has been shared among all ftrace-based kprobes, and checked > when registering. So what we need is to introduce a new kprobe flag which > states that this kprobe doesn't modify regs->ip. And kprobe prepare 2 > ft

Re: kprobes, livepatch and FTRACE_OPS_FL_IPMODIFY

2019-07-24 Thread Masami Hiramatsu
On Wed, 24 Jul 2019 11:19:42 -0400 Joe Lawrence wrote: > Hi Masami, > > I wanted to revisit FTRACE_OPS_FL_IPMODIFY blocking of kprobes and > livepatch, at least in cases where kprobe pre_handlers don't modify > regs->ip. OK, now I think we can pass a flag to kprobe_register() to modify regs->ip

kprobes, livepatch and FTRACE_OPS_FL_IPMODIFY

2019-07-24 Thread Joe Lawrence
Hi Masami, I wanted to revisit FTRACE_OPS_FL_IPMODIFY blocking of kprobes and livepatch, at least in cases where kprobe pre_handlers don't modify regs->ip. (We've discussed this previously at part of a kpatch github issue #47: https://github.com/dynup/kpatch/issues/47) The particular use case I