Re: [PATCH v3 18/26] objtool: Fix !CFI insn_state propagation

[Josh Poimboeuf]

On Tue, Mar 24, 2020 at 04:31:31PM +0100, Peter Zijlstra wrote:
> Objtool keeps per instruction CFI state in struct insn_state and will
> save/restore this where required. However, insn_state has grown some
> !CFI state, and this must not be saved/restored (and thus lost).
> 
> Fix this by explicitly preserving the !CFI state and clarify by
> restucturing the code and adding a comment.
> 
> XXX, the insn==first condition is not handled right.
> 
> Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> ---
>  tools/objtool/check.c |   95 +++++++++++++++++++++++++++++---------------------
>  tools/objtool/check.h |    8 ++++
>  2 files changed, 64 insertions(+), 39 deletions(-)
> 
> --- a/tools/objtool/check.c
> +++ b/tools/objtool/check.c
> @@ -2033,6 +2033,59 @@ static int validate_return(struct symbol
>  	return 0;
>  }
>  
> +static int apply_insn_hint(struct objtool_file *file, struct section *sec,
> +			   struct symbol *func, struct instruction *first,
> +			   struct instruction *insn, struct insn_state *state)
> +{
> +	struct insn_state old = *state;
> +
> +	if (insn->restore) {
> +		struct instruction *save_insn, *i;
> +
> +		i = insn;
> +		save_insn = NULL;
> +		sym_for_each_insn_continue_reverse(file, func, i) {
> +			if (i->save) {
> +				save_insn = i;
> +				break;
> +			}
> +		}
> +
> +		if (!save_insn) {
> +			WARN_FUNC("no corresponding CFI save for CFI restore",
> +					sec, insn->offset);
> +			return 1;
> +		}
> +
> +		if (!save_insn->visited) {
> +			/*
> +			 * Oops, no state to copy yet.
> +			 * Hopefully we can reach this
> +			 * instruction from another branch
> +			 * after the save insn has been
> +			 * visited.
> +			 */
> +			if (insn == first)
> +				return 0; // XXX

Yeah, moving this code out to apply_insn_hint() seems like a nice idea,
but it wouldn't be worth it if it breaks this case.  TBH I don't
remember if this check was for a real-world case.  Might be worth
looking at...  If this case doesn't exist in reality then we could just
remove this check altogether.

> +
> +			WARN_FUNC("objtool isn't smart enough to handle this CFI save/restore combo",
> +					sec, insn->offset);
> +			return 1;
> +		}
> +
> +		insn->state = save_insn->state;
> +	}
> +
> +	*state = insn->state;

This would have been easier to review if apply_insn_hint() were added in
a separate patch.

> +
> +	/* restore !CFI state */
> +	state->df = old.df;
> +	state->uaccess = old.uaccess;
> +	state->uaccess_stack = old.uaccess_stack;

Maybe we should just move the CFI stuff into a state->cfi substruct.
That would remove the need for these bits and probably also the comment
above the insn_state declaration.

-- 
Josh