[PATCH] arm64: ftrace: stop using kstop_machine to enable/disable tracing

[Will Deacon]

On Thu, Dec 03, 2015 at 05:21:22PM +0800, libin wrote:
> 
> on 2015/12/2 20:36, Will Deacon wrote:
> > On Sat, Nov 28, 2015 at 03:50:09PM +0800, Li Bin wrote:
> >> On arm64, kstop_machine which is hugely disruptive to a running
> >> system is not needed to convert nops to ftrace calls or back,
> >> because that modifed code is a single 32bit instructions which
> >> is impossible to cross cache (or page) boundaries, and the used str
> >> instruction is single-copy atomic.
> > This commit message is misleading, since the single-copy atomicity
> > guarantees don't apply to the instruction-side. Instead, the architecture
> > calls out a handful of safe instructions in "Concurrent modification and
> > execution of instructions".
> 
> Right, thank you for your comments.
> 
> > Now, those safe instructions *do* include NOP, B and BL, so that should
> > be sufficient for ftrace provided that we don't patch condition codes
> > (and I don't think we do).
> 
> Yes, and so far this assumption has no probem, but in order to avoid exceeding these
> safe insturctions in the future, we can use aarch64_insn_hotpatch_safe() to verify the
> instruction to determine whether needs stop_machine() to synchronize or use
> aarch64_insn_patch_text directly. Right or I am missing something?

I think you're missing the case where the instruction changes under our
feet after we've read it but before we've replaced it (e.g. due to module
unloading). I think that's why ftrace_modify_code has the comment about
lack of locking thanks to stop_machine.

Will