On 08/24, Will Deacon wrote: > > On Wed, Aug 24, 2016 at 05:47:11PM +0200, Oleg Nesterov wrote: > > On 08/24, Pratyush Anand wrote: > > > > > > > I don't think we want user_{enable,disable{_single_step in the long > > > > term, > > > > please look at 9bd1190a11c9d2 "uprobes/x86: Do not (ab)use > > > > TIF_SINGLESTEP > > > > /user_*_single_step() for single-stepping". it seems that ARM64 > > > > sets/clears > > > > TIF_SINGLESTEP. You can also lool at saved_tf logic, probably ARM64 > > > > needs > > > > the same. > > > > > > IIUC, then you mean that TIF_SINGLESTEP is a per task flag, > > > > Yes, and nobody but ptrace should use it, otherwise ptrace/uprobes can > > confuse > > each other. And uprobes simply doesn't need to set/clear it. > > We're already using it for kprobes, hw_breakpoint and kgdb as well as > ptrace, so I'd rather uprobes either followed existing practice, or we > converted everybody off the current code.
And perhaps this is fine for arm64, I do not know. > In what way do things get confused? Say, arch_uprobe_post_xol() should not blindly do user_disable_single_step(), this can confuse ptrace if TIF_SINGLESTEP was set by debugger which wants to step over the probed insn. > > I can't really answer since I know nothing about arm. x86 just needs to set > > X86_EFLAGS_TF, I guess arm needs to modify some register too? > > We have {user,kernel}_{enable,disable}_single_step for managing the various > registers controlling the single-step state machine on arm64. Yes, and perhaps uprobes can just do set_regs_spsr_ss() ? I never looked into arch/arm64/, but it seems that we only need to ensure that call_step_hook() will be called even if user_mode() == T, why do we need TIF_SINGLESTEP ? Nevermind. I can be easily wrong and let me repeat that I agree with user_{enable,disable}_single_step in the initial version in any case. Oleg.